JP2006065844A - File system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a file system capable of supporting a user who requests various reproduction speeds by improving the use efficiency of a video server and providing an image in a short response time to a request from the user. <P>SOLUTION: When a get resource message is arrived from the outside, a resource assignment part 102 performs reservation of a resource. At the time of reading and writing data, a parameter is checked in an application value management part 111, and assignment of resources is performed in a slot assignment part 103, whereby an efficient file system ensuring a delay quality for reading of data is constructed. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a file system, and more particularly to a file system that handles reading and writing of multimedia data such as video, audio, and moving images that require real-time performance.

  A file system used in a conventional computer has mainly handled text data. For example, according to “UNIX kernel design” (Murice J. Bach / Fumi Sakamoto, Yoshikatsu Tada, Jun Murai; Kyoritsu Publishing Co., Ltd.), the UNIX (registered trademark) file system reads data from a file, Alternatively, when data is written to a file, the file is opened using an open system call, and the contents of the file are read or written using system calls such as read and write.

  An example of a file system is a video server that handles digitized video data. Such video servers include a configuration in which a plurality of storage devices storing video data are connected by an exchange device as described in “Video Server” of Japanese Patent Application Laid-Open No. 07-107425. The video playback procedure in the video server is described in ISO / IEC standard 13818-6 (Committee Draft). According to this Committee Draft, when the user sends “Directory Open” to the video server, the video server completes preparation for continuous playback of the video and sends “DSM Stream Play” “DSM Pause Resume”. Video data transmission to the user starts. In addition, when the supply of video data is stopped, “DSM Stream Pause” is transmitted from the user to the video server.

  However, in a conventional file system such as UNIX (registered trademark), when a large number of accesses from the user occur simultaneously, the read waiting time on the user side increases. Thus, the waiting time from when the user issues a read command to when the reception of data is completed is unpredictable because it is affected by the access status from other users. In particular, like a video server, the file system manages data such as video and audio that require continuity in time, and the user reads out part of the data at regular intervals and plays it on the user terminal. When performing the above, it is necessary to reliably acquire data within a certain time, and the conventional file system could not cope. Therefore, for data that requires continuity in time, such as video data, it is necessary to acquire long data in advance on the user side, and to reproduce the video after acquiring all the data. In this case, it is necessary to hold a large-capacity storage device capable of storing long data on the user side, and a long waiting time is required until the start of video reproduction.

  Therefore, an object of the present invention is to provide a file system that can provide data with a quick response time to a request from a user and can support a user who requests various reproduction speeds.

A first invention is a scheduling device for scheduling service requests so that services are periodically provided from a server to a plurality of clients, respectively.
First queuing means for queuing service requests from a plurality of clients;
A registration means for registering a cycle for generating an execution right for each of a plurality of clients;
An execution right generating means for generating an execution right for executing the service request in relation to the period registered in the registration means;
Execution instruction means for instructing the server to execute a service request from a client corresponding to the execution right generated by the execution right generation means among the service requests queued in the first queue means. Yes.

  As described above, the first queuing means queues the service request from each client. The registering unit registers a cycle for generating an execution right for each client. The execution right generation means generates an execution right in relation to the period registered in the registration means. The execution instructing unit instructs the server to execute a service request from the client corresponding to the execution right generated by the execution right generating unit among the service requests queued in the first queue unit.

  According to a second invention, in the first invention, the execution right generation means is activated at a predetermined cycle, and at the time of activation, the cycle in which the execution right is generated and the cycle in which the execution right is activated are registered. Based on the above, it is determined for each client whether or not the execution right is generated, and the execution right of the client determined to generate the execution right is generated in a lump.

  In a third aspect based on the second aspect, the execution right generation means is activated based on a period in which the execution right is to be generated and a period in which the execution right is activated, registered in the registration means. To determine whether to generate an execution right by calculating the time until the next execution right is generated, and comparing the calculated time with the period at which the self is activated It is characterized by.

  In a fourth aspect based on the third aspect, the execution right generation means is configured such that the time Tofs from the current activation time to the time of generating the next execution right is equal to the cycle Tslt at which the self is activated. The execution right of the client with Tofs <Tslt is generated, and the period Treq to be generated with the execution right registered in the registration unit is added to the Tofs of the client, and the Tofs obtained by the addition is Tofs ≧ In the case of Tslt, a new Tofs is calculated by the formula Tofs = Tofs−Tslt with respect to the Tofs, and the Tofs for the next activation is used.

  In a fifth aspect based on the fourth aspect, the execution right generation means further generates an execution right and adds Treq to the Tofs when the Tofs obtained by the addition satisfies Tofs <Tslt. Then, when the Tofs obtained by repeating and adding the Tslt is equal to Tofs ≧ Tslt, a new Tofs is calculated for the Tofs by the expression Tofs = Tofs−Tslt, and the next time. It is characterized by being Tofs when activated.

  In a sixth aspect based on the fourth aspect, the execution right generation means is configured such that the time Tofs from the current activation time to the time of generating the next execution right is Tofs ≧ Tslt, or the service request However, for a client that is not queued in the first queuing means, the execution right is not generated, and a new Tofs is calculated for the Tofs by the expression Tofs = Tofs-Tslt, and the next activation is performed. It is characterized in that it is Tofs at the time of being performed.

  In a seventh aspect based on the sixth aspect, the execution right generation means sets the Tofs at the next activation to 0 when the new Tofs obtained by the calculation is Tofs <0. It is a feature.

  In an eighth aspect based on the second aspect, the execution instructing means is activated at a predetermined cycle, and execution of the client with a shorter time from the current activation time to the time when the next execution right is generated is performed. It is characterized by selecting rights.

In a ninth aspect based on the eighth aspect, the execution right generation means calculates and generates the evaluation index indicating the priority as a function of the period in which the execution right should be generated, registered in the registration means. Grant to execute rights,
The execution instructing means is characterized by selecting an execution right to be prioritized among the execution rights generated by the execution right generation means based on the assigned evaluation index.

In a tenth aspect based on the ninth aspect, the execution right generation means sets the evaluation index Nsft to a period Treq for generating the execution right registered in the registration means and a period Tslt at which the execution right is started. , Function Nsft = f (Treq / Tslt) (where f (x) is the smallest integer equal to or greater than x),
The execution instruction means selects an execution right with a smaller evaluation index Nsft.

  An eleventh invention is characterized in that, in the tenth invention, the execution right generation means is activated at a cycle shorter than the longest of the cycles registered in the registration means.

  According to a twelfth aspect, in the tenth aspect, the execution right generation unit is activated at a cycle shorter than the shortest of the cycles registered in the registration unit.

  In a thirteenth aspect based on the ninth aspect, the execution instructing means queues the execution right generated by the execution right generation means for each assigned evaluation index and in the order of generation. Among the execution rights queued in the queuing means, the granted evaluation index is smaller and the one generated earlier is selected.

  In a fourteenth aspect based on the ninth aspect, the execution instructing means recognizes the maximum number of service requests to instruct execution at the time of activation, and at the time of instructing execution of service requests by a number equal to the maximum number. The execution instruction is stopped, and the evaluation index of the remaining execution right is changed to a higher priority index.

  In a fifteenth aspect based on the fourteenth aspect, the maximum number of service requests instructed to be executed at the time of activation recognized by the execution instructing means can be executed by the server within a time equal to a period in which the self is activated. It is characterized by the maximum number of service requests.

  In a sixteenth aspect based on the tenth aspect, when the execution instructing unit recognizes the maximum number of service requests to instruct execution at the time of activation and instructs execution of service requests by the number equal to the maximum number. The execution instruction is stopped, and the evaluation index of the remaining execution right is decremented.

  In a seventeenth aspect based on the sixteenth aspect, the maximum number of service requests that are instructed to be executed at the time of activation recognized by the execution instructing means can be executed by the server within a period equal to a period in which the server is activated. It is characterized by the maximum number of service requests.

  An eighteenth aspect of the invention is characterized in that, in the fourteenth aspect, the information processing apparatus further comprises registration instruction means for instructing the registration means to register a period for generating an execution right based on declarations from a plurality of clients. .

  In a nineteenth aspect based on the eighteenth aspect, the registration instruction means recognizes the maximum number of service requests to be executed by the execution instruction means per unit time, and in addition to the already registered period, When it is declared that another period is registered, if it is determined that the average number of execution rights generated per unit time does not exceed the maximum number even if it is further registered, the registration means is instructed. The period is registered, and if it is further registered, if it is determined that the period is exceeded, the period is not registered.

A twentieth aspect of the invention is a scheduling apparatus that schedules service requests so that services are periodically provided from a plurality of servers to clients.
First queue means for queuing service requests to a plurality of servers;
Registration means for registering a cycle for generating an execution right for each of a plurality of servers;
An execution right generating means for generating an execution right for executing the service request in relation to the period registered in the registration means;
Execution instruction means for instructing the server to execute a service request to the server corresponding to the execution right generated by the execution right generation means among the service requests queued in the first queue means ing.

  As described above, the first queuing means queues a service request to each server. The registering unit registers a cycle for generating an execution right for each server. The execution right generation means generates an execution right in relation to the period registered in the registration means. The execution instructing unit instructs the server to execute a service request of the server corresponding to the execution right generated by the execution right generating unit among the service requests queued in the first queue unit.

  According to a twenty-first aspect, in the twentieth aspect, the execution right generation means is activated at a predetermined cycle, and at the time of activation, the cycle in which the execution right is to be generated and the cycle in which the execution right is activated Based on the above, it is determined for each server whether or not the execution right is to be generated, and the execution right of the server that is determined to generate the execution right is collectively generated.

  According to a twenty-second aspect, in the twenty-first aspect, the execution right generation means is activated based on a period in which the execution right is to be generated and a period in which the execution right is activated, registered in the registration means. To determine whether to generate an execution right by calculating the time until the next execution right is generated, and comparing the calculated time with the period at which the self is activated It is characterized by.

  According to a twenty-third aspect, in the twenty-second aspect, the execution right generation means is configured such that a time Tofs from the current activation time to a time for generating the next execution right is equal to a cycle Tslt in which the self is activated. The execution right of the server where Tofs <Tslt is generated, and the period Ttv for generating the execution right registered in the registration means is added to the Tofs of the server, and the Tofs obtained by the addition is Tofs ≧ In the case of Tslt, a new Tofs is calculated by the formula Tofs = Tofs−Tslt with respect to the Tofs, and the Tofs for the next activation is used.

  In a twenty-fourth aspect based on the twenty-third aspect, the execution right generation means further generates an execution right and adds Titv to the Tofs when the Tofs obtained by addition is Tofs <Tslt. Then, when the Tofs obtained by repeating and adding the Tslt is equal to Tofs ≧ Tslt, a new Tofs is calculated for the Tofs by the expression Tofs = Tofs−Tslt, and the next time. It is characterized by being Tofs when activated.

  In a twenty-fifth aspect based on the twenty-third aspect, the execution right generation means is configured so that a time Tofs from a time when the execution right is activated to a time when the next execution right is generated is Tofs ≧ Tslt or a service request However, for a server that is not queued in the first queuing means, the execution right is not generated, and a new Tofs is calculated for the Tofs by the formula Tofs = Tofs-Tslt, and the next activation is performed. It is characterized in that it is Tofs at the time of being performed.

  In a twenty-sixth aspect based on the twenty-fifth aspect, the execution right generation means sets the Tofs at the next activation to 0 when the new Tofs obtained by calculation is Tofs <0. It is a feature.

  In a twenty-seventh aspect based on the twenty-first aspect, the execution instructing means is activated at a predetermined cycle, and the time from the activated current time to the time when the next execution right is generated is shorter. It is characterized by selecting rights.

In a twenty-eighth aspect based on the twenty-seventh aspect, the execution right generating means calculates and generates the evaluation index indicating the priority as a function of a cycle in which the execution right is to be generated registered in the registration means. Grant to execute rights,
The execution instructing means is characterized by selecting an execution right to be prioritized among the execution rights generated by the execution right generation means based on the assigned evaluation index.

In a twenty-ninth aspect based on the twenty-eighth aspect, the execution right generation means sets the evaluation index Nsft to a period Ttv for generating the execution right registered in the registration means and a period Tslt in which the execution right is activated. Nsft = f (Titv / Tslt) (where f (x) is the smallest integer equal to or greater than x),
The execution instruction means selects an execution right with a smaller evaluation index Nsft.

  A thirtieth aspect of the present invention is characterized in that, in the twenty-ninth aspect, the execution right generation means is activated with a period shorter than the longest period registered in the registration means.

  According to a thirty-first aspect, in the twenty-ninth aspect, the execution right generation means is activated at a cycle shorter than the shortest of the cycles registered in the registration means.

  In a thirty-second aspect based on the twenty-eighth aspect, the execution instruction means queues the execution right generated by the execution right generation means for each assigned evaluation index and in the order of generation. Among the execution rights queued in the queuing means, the granted evaluation index is smaller and the one generated earlier is selected.

  In a thirty-third aspect based on the twenty-eighth aspect, when the execution instructing unit recognizes the maximum number of service requests to instruct execution at the time of activation and instructs execution of service requests by the number equal to the maximum number. The execution instruction is stopped, and the evaluation index of the remaining execution right is changed to a higher priority index.

  In a thirty-fourth aspect, in the thirty-third aspect, the maximum number of service requests instructed to be executed at the time of activation recognized by the execution instructing means is from all servers to the client within a time equal to a period in which the self is activated. The number of service requests corresponding to the maximum amount of information that can be transmitted.

  In a thirty-fifth aspect based on the twenty-ninth aspect, when the execution instructing unit recognizes the maximum number of service requests to instruct execution at the time of activation and instructs execution of service requests by the number equal to the maximum number. The execution instruction is stopped, and the evaluation index of the remaining execution right is decremented.

  According to a thirty-sixth aspect, in the thirty-fifth aspect, the maximum number of service requests that are recognized by the execution instructing means at the time of start-up The number of service requests corresponding to the maximum amount of information that can be transmitted.

  According to a thirty-seventh aspect, in the thirty-third aspect, the information processing apparatus further includes registration instruction means for instructing the registration means to register / update a cycle for generating the execution right based on a report from the client. .

  In a thirty-eighth aspect based on the thirty-seventh aspect, the registration instructing means recognizes the maximum number of service requests that the execution instructing means instructs all servers to execute per unit time, and the period is already registered. In addition, when it is declared to register the cycle of another server, the average number of execution rights per unit time of all servers that are generated will be determined not to exceed the maximum number even if it is registered. In this case, the registration means is instructed to register the period, and if it is further registered, if it is determined that the period is exceeded, the period is not registered.

  In a thirty-ninth aspect based on the thirty-seventh aspect, the registration instructing means recognizes the maximum number of service requests that the execution instructing means instructs each server to execute per unit time, and the period is already registered. In addition, when the declaration of registering the cycle of another server is made, the average number of execution rights of the server to be generated per unit time generated by the execution instruction means per unit time is If it is determined that the maximum number of service requests for instructing execution to the server is not exceeded, the registration means is instructed to register the cycle, and if it is further registered, if it is determined that it will be exceeded, it will not be registered. It is a feature.

  In a fortieth aspect, in the thirty-ninth aspect, the maximum number of service requests that the execution instruction unit instructs each server to execute per unit time, recognized by the registration instruction unit, is determined by each server per unit time. It is characterized by the maximum number of service requests that can be executed.

  In a forty-first aspect according to the thirty-seventh aspect, the registration instruction means includes a maximum number of service requests that the execution instruction means instructs all servers to execute per unit time, and the execution instruction means per unit time that each execution instruction means sends to each server. Recognizes the maximum number of service requests to instruct execution, and updates the already registered period Ttv for generating the right of execution of a server to one that further adds the period Treq to generate the right of execution The average number of execution rights of all servers generated per unit time, even if it is updated when a report is received, is the number of service requests that the execution instruction means instructs all servers to execute per unit time. It is determined that the maximum number is not exceeded, and the average number of execution rights of the server that are generated per unit time is executed by the execution instruction unit per unit time. When it is determined that the maximum number of service requests to be instructed is not exceeded, the registration unit is instructed to change the already registered Titv to a new Titv calculated by the expression Titv = 1 / (1 / Titv + 1 / Treq). It is characterized in that it is updated, and if it is updated, it is not updated when it is determined that it will be exceeded.

  In a forty-second aspect based on the thirty-seventh aspect, the registration instructing means updates the already-registered period Titv for generating the right of execution of a server to one that generates the right of execution excluding the period Treq. When the report is received, the registration means is instructed to update the already registered Titv to a new Titv calculated by the expression Titv = 1 / (1 / Titv-1 / Treq). Yes.

  According to the first aspect of the invention, while queuing the service request from each client, the cycle for generating the execution right for each client is registered, and the execution right is generated based on the cycle. . Then, the server requests the server to execute the queued service request from the client corresponding to the generated execution right. As a result, the time from when a service request from one client is sent to when it is executed is not affected by the status of the transmission of the service request from another client, so the transmitted service request is executed. It is possible to prevent indefinite time until it is done.

  According to the second aspect, the execution right generation means is activated at a predetermined cycle and performs an operation of generating an execution right. At this time, the execution right generation means determines whether to generate the execution right for each client based on the cycle in which the execution right should be generated and the cycle in which the execution right is activated, and determines that the execution right is generated. The execution right of the client is generated in a lump. As a result, the processing operation of the execution right generation means can be reduced as compared with the case where the execution right is generated in the cycle in which the execution right should be generated and registered in the registration means.

  According to the third aspect, it is possible to determine whether or not to generate an execution right for each client.

  According to the fourth aspect of the invention, the execution right generation means generates the execution right of the client whose Tofs is Tofs <Tslt with respect to Tslt. Further, Treq is added to the Tofs of the client, and when the Tofs obtained by the addition is Tofs ≧ Tslt, a new Tofs is calculated for the Tofs by the formula Tofs = Tofs−Tslt, and is calculated. The Tofs obtained in this way are taken as Tofs for the next activation. Thereby, it is possible to determine the client that should generate the execution right and generate the execution right.

  According to the fifth aspect, the execution right generation means further generates the execution right when the Tofs obtained by the addition is Tofs <Tslt, and adds the Treq to the Tofs and adds the Treq to the Tslt. Repeat the operation to compare. Then, when the Tofs obtained by the addition becomes Tofs ≧ Tslt, a new Tofs is calculated with respect to the Tofs by the expression Tofs = Tofs−Tslt, and is set as the Tofs for the next activation. . As a result, it is possible to further determine the clients that should generate a plurality of execution rights and generate the necessary number of execution rights.

  According to the sixth invention, the execution right generation means generates an execution right for a client whose Tofs is Tofs ≧ Tslt or whose service request is not queued in the first queue means. I won't let you. Then, for the Tofs, a new Tofs is calculated by the formula Tofs = Tofs−Tslt and set as Tofs for the next activation. As described above, by not generating an execution right for a client for which a service request is not queued, it is possible to prevent a service from being provided to a client that has not transmitted a service request.

  According to the seventh aspect, the Tofs of a client for which a service request is not queued is repeatedly subtracted from Tslt to become a small value, and when the service request is queued, many execution rights for the client are generated. Can be prevented.

  According to the eighth aspect of the invention, by selecting a client execution right that has a shorter time from the current activation time to the time when the next execution right is generated, the number of delays generated can be further increased. Can be reduced.

  According to the ninth aspect, the execution right generation means assigns the evaluation index indicating the priority calculated as a function of the cycle for generating the execution right to the generated execution right, and the execution instruction means includes: Based on the assigned evaluation index, the execution right to be prioritized is selected. Thereby, when the execution rights compete, it is possible to select the one that should be prioritized in relation to the cycle in which the execution rights are to be generated.

  According to the tenth aspect of the present invention, when execution rights compete, it is possible to select one with a shorter cycle in which the execution right should be generated.

  According to the eleventh aspect, the same evaluation index is not given to the execution rights of all clients.

  According to the twelfth aspect, a more detailed evaluation index is assigned to the execution right of each client.

  According to the thirteenth aspect, the execution instructing unit can easily select a priority among the generated execution rights, and therefore the processing operation of the execution instructing unit can be reduced.

  According to the fourteenth aspect, the execution instruction means stops the execution instruction when the maximum number of execution instructions are given, and sets the remaining execution right evaluation index to a higher priority. change. Thereby, it is possible to prevent the execution instruction from being performed beyond the maximum number, and it is possible to prevent the remaining execution right from being selected for a long time.

  According to the fifteenth aspect of the present invention, it is possible to prevent the execution instruction from being performed beyond the processing capacity of the server.

  According to the sixteenth aspect, when the maximum number of execution instructions are issued, the execution instruction means stops the execution instruction and decrements the remaining execution right evaluation index. As a result, it is possible to prevent the execution instruction from being performed beyond the maximum number, and the evaluation index of the remaining execution right is decremented every Tslt, so that it is generated until Treq + Tslt at the latest. In addition, the execution right can be selected.

  According to the seventeenth aspect of the present invention, it is possible to prevent the execution instruction from being performed beyond the processing capacity of the server.

  According to the eighteenth aspect, the period for generating the execution right can be registered in the registration means as necessary.

  According to the nineteenth aspect of the invention, the registration instructing means calculates the average per unit time of the execution right generated by the execution right generation means even if the declared period is further registered in addition to the already registered period. Only when it is determined that the number does not exceed the maximum number of service requests to be executed by the execution instruction unit per unit time, the registration unit is instructed to register the declared cycle. Thus, when a new cycle is registered, the average number of execution rights generated per unit time can be prevented from exceeding the maximum number of service requests instructed to be executed by the execution instruction unit per unit time. .

  According to the twentieth invention, while queuing a service request to each server, a cycle for generating an execution right for each server is registered, and the execution right is generated based on the cycle. . Then, among the queued service requests, the server requests corresponding to the generated execution right are instructed to be executed. As a result, the time from when a service request to one server is sent to when it is executed is not affected by the status of sending the service request to another server, so the sent service request is executed. It is possible to prevent indefinite time until it is done.

  According to the twenty-first aspect, the execution right generation means is activated at a predetermined cycle and performs an operation of generating an execution right. At this time, the execution right generation means determines whether to generate the execution right for each server based on the cycle in which the execution right should be generated and the cycle in which the execution right is activated, and determines that the execution right is generated. The server execution right is generated in a batch. As a result, the processing operation of the execution right generation means can be reduced as compared with the case where the execution right is generated in the cycle in which the execution right should be generated and registered in the registration means.

  According to the twenty-second aspect, it is possible to determine whether or not to generate an execution right for each server.

  According to the twenty-third aspect, the execution right generation means generates the execution right of the server whose Tofs is Toss <Tslt with respect to Tslt. Further, when Ttv is added to the Tofs of the server, and the Tofs obtained by the addition is Tofs ≧ Tslt, a new Tofs is calculated for the Tofs by the formula Tofs = Tofs−Tslt. The Tofs obtained in this way are taken as Tofs for the next activation. Thereby, it is possible to determine the server that should generate the execution right and generate the execution right.

  According to the twenty-fourth aspect, the execution right generation means further generates an execution right when the Tofs obtained by the addition is Tofs <Tslt, and adds Ttv to the Tofs and adds the Ttv to the Tslt. Repeat the operation to compare. Then, when the Tofs obtained by the addition becomes Tofs ≧ Tslt, a new Tofs is calculated with respect to the Tofs by the expression Tofs = Tofs−Tslt, and is set as the Tofs for the next activation. . Accordingly, it is possible to further determine a server that should generate a plurality of execution rights and generate a necessary number of execution rights.

  According to the twenty-fifth aspect, the execution right generation means generates an execution right for a server whose Tofs is Tofs ≧ Tslt or the service request is not queued in the first queue means. I won't let you. Then, for the Tofs, a new Tofs is calculated by the formula Tofs = Tofs−Tslt and set as Tofs for the next activation. As described above, by not generating an execution right for a server for which a service request is not queued, it is possible to prevent a server that has not received a service request from providing a service.

  According to the twenty-sixth aspect of the present invention, Tofs of a server for which a service request is not queued is repeatedly subtracted from Tslt to become a small value, and when the service request is queued, many execution rights for the server are generated. Can be prevented.

  According to the twenty-seventh aspect, by selecting a server execution right that has a shorter time from the current activation time to the time when the next execution right is generated, the number of delays can be increased. Can be reduced.

  According to the twenty-eighth aspect, the execution right generation means assigns the evaluation index indicating the priority calculated as a function of the cycle in which the execution right should be generated to the generated execution right, and the execution instruction means includes: Based on the assigned evaluation index, the execution right to be prioritized is selected. Thereby, when the execution rights compete, it is possible to select the one that should be prioritized in relation to the cycle in which the execution rights are to be generated.

  According to the twenty-ninth aspect, when execution rights compete, it is possible to select a shorter cycle for generating execution rights.

  According to the thirtieth aspect, the same evaluation index is not given to the execution rights of all servers.

  According to the thirty-first aspect, a more detailed evaluation index is given to the execution right of each server.

  According to the thirty-second aspect, the execution instructing unit can easily select a priority among the generated execution rights, and therefore the processing operation of the execution instructing unit can be reduced.

  According to the thirty-third aspect, the execution instruction means stops the execution instruction when the maximum number of execution instructions are given, and sets the remaining execution right evaluation index to a higher priority. change. Thereby, it is possible to prevent the execution instruction from being performed beyond the maximum number, and it is possible to prevent the remaining execution right from being selected for a long time.

  According to the thirty-fourth aspect, execution instructions that exceed the maximum amount of information that can be transmitted from all servers to the client can be prevented.

  According to the thirty-fifth aspect, the execution instruction means stops the execution instruction and decrements the remaining execution right evaluation index when the maximum number of execution instructions are issued. As a result, it is possible to prevent the execution instruction from being performed beyond the maximum number, and the evaluation index of the remaining execution right is decremented every Tslt, so that it is generated until Treq + Tslt at the latest. In addition, the execution right can be selected.

  According to the thirty-sixth aspect, it is possible to prevent a number of execution instructions from exceeding the maximum amount of information that can be transmitted from all servers to the client.

  According to the thirty-seventh aspect, the period for generating the execution right can be registered / updated by the registration unit as necessary.

  According to the thirty-eighth aspect, the registration instruction means is a unit of execution rights of all servers generated by the execution right generation means even if the declared period is further registered in addition to the already registered periods. Only when it is determined that the average number per hour does not exceed the maximum number of service requests that the execution instruction means instructs execution to all servers per unit time, the registration means is instructed to register the declared cycle. Let As a result, when the declared server cycle is registered, the average number of execution rights per unit time of all servers generated is the number of service requests that the execution instruction means instructs all servers to execute per unit time. It is possible to prevent the maximum number from being exceeded.

  According to the thirty-ninth aspect of the invention, the registration instructing means generates the execution right of the server, which is generated by the execution right generating means even if the declared server period is further registered in addition to the already registered period. Only when it is determined that the average number per unit time does not exceed the maximum number of service requests that the execution instruction means instructs the server to execute per unit time, the registration means is instructed and declared To register. As a result, when registering the declared server cycle, the average number of execution rights of the server per unit time generated is the number of service requests that the execution instruction means instructs the server to execute per unit time. It is possible to prevent the maximum number from being exceeded.

  According to the fortieth aspect, the service request that can be executed by the server per unit time is the average number of execution rights of the server generated per unit time that is generated when the declared server cycle is registered. Can be prevented from exceeding the maximum number.

  According to the forty-first aspect of the present invention, when a report is received so as to update the Titv of a server to one that generates an execution right by adding a cycle Treq, The average number of execution rights per unit time does not exceed the maximum number of service requests that the execution instruction means instructs all servers to execute per unit time, and is generated per unit time of the execution rights of the server. Only when it is determined that the average number does not exceed the maximum number of service requests that the execution instruction means instructs the server to execute per unit time, the registration means is instructed to update Titv of the server. As a result, when the Titv of the server is updated, the average number of execution rights of all servers per unit time is the maximum number of service requests that the execution instruction unit instructs all servers to execute per unit time. The average number of execution rights of the server that are generated per unit time is the maximum number of service requests that the execution instruction means instructs the server to execute per unit time. Can not exceed.

  According to the forty-second aspect of the present invention, when a declaration is received to update the already registered cycle for generating the execution right to one that is generated excluding the cycle Treq, the already registered cycle is You can update to a new one.

(First embodiment)
FIG. 1 is a block diagram showing a configuration of a file system according to the first embodiment of the present invention. In FIG. 1, the file system includes an exchange device 1, a file manager 2, and a plurality of server units 3.

  The file manager 2 manages the entire file system. Each server unit 3 accepts a file read or write request from the user, and transmits data or a message read from a storage device managed by the file system to the user. Data is exchanged between each server unit 3 and between each server unit 3 and the file manager 2 via the exchange device 1. Data managed by the file system is stored in a storage device existing in each server unit 3.

  FIG. 2 is a block diagram showing a configuration of a part related to file access in each server unit 3. In FIG. 2, each server unit 3 includes an external message interface unit 101, a resource allocation unit 102, a slot allocation unit 103, an address search unit 104, a request transmission unit 105, an external network I / O control unit 106, , An internal message interface unit 107, a storage device 108, a storage control unit 109, an internal network I / O control unit 110, and a reported value management unit 111.

  The external message interface unit 101 identifies a message exchanged with the outside, and gives a processing instruction to each functional block in the server unit. The resource allocation unit 102 allocates resources in the file system. The slot allocation unit 103 allocates slots managed by the server unit 3. The slot will be described later. The address search unit 104 searches the storage location of the requested data. The request sending unit 105 instructs sending of a block read request for requesting reading of a fixed-length data block and a block write request for requesting writing. The internal message interface unit 107 controls transmission / reception of messages to / from the file manager 2 inside the file system and other server units. The storage device 108 stores data block and data storage location information for each file. The storage control unit 109 controls access to the storage device 108. The internal network I / O control unit 110 controls the timing of sending the data block requested by the block read request to the switching apparatus 1. The declared value management unit 111 checks whether or not the parameters of the message arriving from the outside are valid.

  FIG. 3 is a block diagram showing a configuration of a part related to file access in the file manager 2. In FIG. 3, the file manager 2 includes a message interface unit 201 and a resource management unit 202. The message interface unit 201 identifies a message exchanged with each server unit 3 via the exchange device 1 and gives a processing instruction to a function in the file manager 2. The resource management unit 202 manages the read bandwidth of the storage device and the capacity and usage of the input / output bandwidth of the switching device in the file system.

Here, the following message arrives at the file system in relation to file access from the outside.
(1) GetResource
(2) FileOpen
(3) ReadBlock
(4) WriteBlock

  The operation of the portion related to file access in the file manager 2 and each server unit 3 when the above message is received will be described below.

  When the GetResource message arrives, the external message interface unit 101 in the server unit 3 notifies the resource allocation unit 102 to that effect. The resource allocation unit 102 instructs the internal message interface unit 107 to send a resource allocation request to the file manager 2. When the resource allocation request arrives at the file manager 2, the resource management unit 202 determines whether the resource allocation request can be accepted and sends the result to the server unit 3. When the resource allocation unit 102 in the server unit 3 receives the determination result indicating that allocation is possible, the resource allocation unit 102 reserves the resource.

  When the FileOpen message arrives, the server unit 3 acquires the storage location information of the data in the file from the storage device 108, and shifts to a file accessible state.

  When the ReadBlock message arrives, the slot allocation unit 103 in the server unit 3 allocates the reserved resource and reads the requested data.

  When the WriteBlock message arrives, the slot allocation unit 103 in the server unit 3 allocates the reserved resource and writes the requested data.

  FIG. 4 is a diagram showing the relationship between resource reservation and allocation and data read operations. Hereinafter, the operation of the file system that performs resource reservation and allocation to read and write data will be described with reference to FIG. First, the MS arrival slot and write slot used for resource reservation and allocation will be described, and then the operation of the file system will be described.

(1) MS Arrival Slot In this file system, the concept of “MS arrival slot” divided every several tens of milliseconds is used. In FIG. 4, a line 801 indicates that the time is divided into MS arrival slots for management. When data is read, the requested data is divided into fixed-length data blocks (MS blocks), and a block read request is sent to the storage device 108 or another server unit for each MS to perform reading. When a server unit sends a block read request to another server unit, an MS arrival slot is allocated. A maximum value is set for the number of block read requests that can be allocated to one slot. In this file system, an MS arrival slot is reserved in advance for a user who has received a block read request in order to guarantee provision of a stream to the user. The MS arrival slot assigned to the block read request is selected from the reserved MS arrival slots. The operation of reserving the MS arrival slot is performed collectively at regular intervals. The parameters forming the “MS arrival slot” are shown below.
T_slt: Slot time interval Nmax1: Number of requests that can be assigned to a slot T_rsvMS: Arrival slot reservation operation interval

  In FIG. 4, reference numeral 802 indicates the interval of T_slt. FIG. 5 shows an example of an MS arrival slot management table for managing the state of the MS arrival slot. As shown in FIG. 5, the MS arrival slot management table allocates “slot number” for identifying a slot, “reserved user number” for identifying presence / absence of a slot and a reserved user, and a block read request. The information includes items such as “assignment request number” for identifying whether or not the request is assigned and “non-priority flag” for identifying whether or not a non-priority block read request may be assigned. A user to which a resource is assigned by the GetResource message uses the reserved slot. In addition, for a user who has requested reading of a block without resource allocation, a slot that can be allocated is allocated with reference to the non-priority flag when reading the block.

(2) Write Slot When writing a block, the concept of “write slot” is used. In each server unit 3, a writing slot is provided for each destination server unit. Data arriving from the outside together with the WriteBlock command is divided into fixed-length data blocks (MS blocks), the destination is determined according to the free area list in the storage device 108, and is sent out as a block write request. Here, a write slot of the destination server unit is always assigned to the block write request. An example of the write slot management table for managing the state of the write slot is shown in FIG. As shown in FIG. 6, the write slot management table includes a “destination unit number” for identifying the destination server unit, a “slot number” for identifying the write slot, the presence / absence of the request assigned to the slot, and the assigned request. Information such as an “assignment request identification number” is identified as information.

(3) Resource Management for Each User In the server unit 3, the users who have received GetResource are managed by “stream management table” in the order of the larger reserved band and the smaller band for each Read / Write. Depending on the reserved bandwidth, the user can read one or more files. FIG. 7 shows an example of the stream management table. This stream management table is managed by the resource allocation unit 102 in each server unit 3. The “stream management table” has items as shown in FIG. 7 as information in order to manage the resource acquisition status for each user. In FIG. 7, “resource descriptor” is an identifier assigned every time a GetResource message is received. The “stream management table” includes, for each “resource descriptor”, “separate read / write” for identifying whether it is for reading or writing, and the average reservation interval of slots necessary for providing the requested bandwidth. Is remembered. Furthermore, the “stream management table” includes “internal bandwidth in the reserved file system” for storing the reserved bandwidth, and “currently usable file system internal bandwidth for storing the amount of bandwidth that is not currently used in the reserved bandwidth”. ”As information.

(4) Resource acquisition acceptance control Resource acquisition acceptance control is control for determining whether or not to accept an external GetResource message and returning the determination result to the outside. Judgment of acceptability is performed by the resource management unit 202 based on the parameters reported by the message. The GetResource message has the following declaration parameters.
Minimum Read / Write interval T of another block of Read / Write
Delay request value D_lim from arrival of ReadBlock to block transmission

  FIG. 8 is a block diagram showing a more detailed configuration of the resource management unit 202 shown in FIG. In FIG. 8, the resource management unit 202 includes a report parameter processing unit 401, a write bandwidth calculation unit 402, an acceptance determination unit 403, and a resource storage unit 404. The report parameter processing unit 401 replaces the parameter reported by the command with a numerical value necessary for acceptance determination. The write band calculation unit 402 calculates a write band. The reception determination unit 403 determines whether messages can be received. The resource storage unit 404 stores the current resource usage status and the maximum value of resources that can be provided.

The reporting parameter processing unit 401 first obtains a value T_max that satisfies the following equation (1) based on the delay request value D_lim.
D_lim = D_net + D_dsk_min + 2 × T_max (1)
In the above equation (1), let D_net be the allowable delay value in the switching apparatus 1, and D_dsk_min be the minimum delay limit value in the storage device 108. Further, when T_max and T are compared, the smaller one is defined as T_min.

  The reporting parameter processing unit 401 calculates a necessary bandwidth amount from the value of T_min. In the case of reading, the reporting parameter processing unit 401 notifies the reception determining unit 403 of the calculated bandwidth amount. In the case of writing, the reporting parameter processing unit 401 adopts the reporting value as the bandwidth of the entrance / exit from the exchange device 1 to the server unit 3, and the reporting value is determined for each storage device as part of the processing capacity of the storage device 108. A numerical value (calculated by the write bandwidth calculation unit 402) multiplied by a certain value is obtained as the peak provision speed. The reception determination unit 403 determines whether the total amount of acquired resources when the GetResource is received is less than or equal to a certain value on the entrance side and the exit side from the exchange device 1 to the server unit 3 and the storage device 108. If reception is possible, the reception determination unit 403 stores the information in the resource storage unit 404 and adds information to the stream management table managed by the resource allocation unit 102. At this time, information is added at a position that does not disturb the description order of the stream management table (from the smallest T_min to the largest).

(5) Reserving resource for reading The MS arrival slot used for the user who has received the GetResource message and has acquired the resource for reading is reserved. The reservation of the MS arrival slot is performed at regular intervals. Reservation of MS arrival slots during the time interval T_rsv is performed in the order described in the stream management table (in the order of the speed of the provided stream from the highest to the lowest). A reservation is made according to the set value of each T_min. In FIG. 4, reference numbers 803 to 806 indicate timings for performing a reservation operation, and reference number 807 indicates a time interval T_rsv. The reservation slot determination method is shown below.

1) For each user, a slot is selected at a time interval of T_min, and each selected slot is set as an expected slot.
2) At timings 803 to 806, an expected reservation slot is determined at a time interval of T_min within the time interval T_rsv.
3) If an appointment slot can be reserved, make a reservation.
4) If the reserved slot cannot be reserved, the next slot after the expected slot is referred to. If the reserved slot can be reserved, it is reserved. If the reservation cannot be made, the next slot is referred. Repeat this.
5) Even when the reserved section is off, a reservation is made for a slot in the next time section T_rsv according to the above rules.

  In FIG. 4, reference numerals 808 to 820 indicate expected slots for reservation. Reference numerals 821 to 833 indicate slots that have actually been reserved. When a new GetResource command is received, slot reservation in the already reserved section is immediately performed according to the set value of T_min. The slot reservation method is the same as that described above. For all users, the non-priority flag in FIG. 5 is set to ON for the MS arrival slot at the time when the reservation operation is performed and which is not reserved.

  Note that the reservation of the MS arrival slot may be performed in order from the earliest arrival timing of the next expected slot. In this case, in the stream management table, information regarding the user who has received the GetResource is described in the order of the arrival timing of the next expected slot. When the MS arrival slot is reserved, the description order of the stream management table is updated.

(6) Block read operation When a ReadBlock message arrives from the outside, the fact is notified to the declared value management unit 111 via the message interface unit 101. The ReadBlock message has the following declaration parameters.
Resource descriptor block read interval Delay request value D_lim from arrival of ReadBlock to block transmission
File identifier Location information in the file

The reported value management unit 111 checks whether or not the block acquisition request interval by the ReadBlock command from the user is as reported by GetResource. The time obtained by adding the delay request value added to the command to the arrival time of ReadBlock is the request arrival time of the block. Check whether the requested arrival time interval of this block is as reported. If there is a violation, a notification of violation is given. Also, it is confirmed that the delay request value is not too small. That is, when the following equation (2) is established, the allowable delay value in the switching device 1 is D_net and the minimum delay limit value in the storage device 108 is D_dsk_min, and a violation is notified to the user.
D_lim <D_net + D_dsk_min + 2 × T_min (2) On the other hand, if it is as reported by GetResource, arrival of ReadBlock is notified to the slot allocation unit 103, the address search unit 104, and the external network I / O control unit 106. The slot allocation unit 103 determines a block read request transmission interval so that data blocks can be continuously provided in the requested bandwidth. By sending a block read request, the interval for acquiring MS blocks (MS arrival interval) is set to be an average interval. When the block read request sending time is reached, MS arrival slot allocation and delay limit value calculation are performed based on the delay request value of each command.

However, slots are not allocated when addressed to the same server unit. The delay limit value D_dsk given to the block read request is defined as the following equation (3).
D_dsk = D_lim (3)

In the case of addressing to another server unit, the slot to be allocated is selected by the following method. The time at which the block read request is sent is T_req. A slot reserved for this user is allocated between times T satisfying the following expression (4). If there are a plurality of reserved slots, the earliest time is selected.
T_req + D_lim−2 × T_min <t ≦ T_req + D_lim (4)

The delay limit value D_dsk to be given to the block read request is obtained by the following equation (5), where t_arv is the time of the assigned slot and D_net is the allowable delay value in the switching apparatus 1.
D_dsk = t_arv−T_req−D_net (5)

  The slot allocation unit 103 instructs the request transmission unit 105 to transmit a block read request using the delay limit value as a parameter. The slot allocation unit 103 also turns on arrival slots that have been reserved but not allocated. In FIG. 4, arrows 834 to 838 indicate the arrival of the ReadBlock message. Reference numerals 839 to 843 indicate MS arrival slots assigned to the respective ReadBlock messages. Arrows 844 to 848 indicate the timing for instructing transmission of a block read request. The request sending unit 105 sends a block read request to the destination declared by the address search unit 104. The internal message interface unit 107 transmits a block read request. In the destination server unit 3, a block read request arrives at the internal network control unit 110 via the internal message interface unit 107. The internal network control unit 110 calculates and stores a time for instructing transmission of the read MS based on the delay limit value given as a parameter. Further, the storage control unit 109 is instructed to read the MS. The storage control unit 109 reads the MS requested from the storage device 108. In the internal network I / O control unit 110, for the MS read from the storage device, the output from the server unit to the switching device is permitted after the time indicated by the delay limit value after the block read request arrives. . The read MS arrives at the external network I / O control unit 106 via the internal message interface unit 107. The external network I / O control unit 106 instructs transmission to the user at time T_req + D_lim. In FIG. 4, reference numbers 849 to 853 indicate the timing at which the MS arrives at the external network I / O control unit 106, and reference numbers 854 to 858 indicate the timing at which the MS is sent to the user.

  For a ReadBlock message that is not specified by the resource descriptor, it is assigned to a slot for which the non-priority flag in FIG.

(7) Block writing operation The stream data arriving from the outside together with the WriteBlock command is checked by the report value management unit 111 to see if it has arrived beyond the bandwidth secured by GetResource. In case of violation, a notification of violation is given. When arriving within the range of the band secured by GetResource, the slot allocation unit 103 divides it into MSs, the address search unit 104 determines the destination, and sends it as a block write request. Here, for the block write request, the slot allocation unit 103 always assigns the write slot of the destination server unit. The block write request is sent at the allocated time.

(Second Embodiment)
FIG. 9 is a block diagram showing a configuration of a file system according to the second embodiment of the present invention. In FIG. 9, the file system includes a file manager 901, a plurality of I / O units 902, a plurality of block storage units 905, and an exchange device 908.

  The file manager 901 manages the entire file system. That is, this file manager 901 has the same function as the file manager 2 shown in FIG. Each I / O unit 902 accepts a file read or write request from the user, and transmits data or a message read from a storage device managed by the file system to the user. Each block storage unit 905 holds a storage device that stores data managed by the file system, and controls reading and writing of data from the storage device. The exchange device 908 connects the file manager 901, each I / O unit 902, and each block storage unit 905, and controls the exchange of data and messages.

  Data managed by this file system is stored in a storage device existing in the block storage unit 905. The configuration of the file access unit in each I / O unit 902 is as shown in FIG. The function of each block shown in FIG. 10 is the same as the function of the block having the same name in FIG. The configuration of the file access unit in each block storage unit 905 is as shown in FIG. The function of each block shown in FIG. 11 also has the same function as the block having the same name in FIG. That is, the I / O unit 902 and the block storage unit 905 cooperate to perform the same function as the server unit 3 in FIG.

(Third embodiment)
FIG. 12 is a block diagram showing a configuration of a file system according to the third embodiment of the present invention. In FIG. 12, the file system includes a message interface unit 1201, a resource management unit 1202, a read / write control unit 1203, a storage device 1204, and a reported value management unit 1205.

  The message interface unit 1201 controls message transmission / reception from the user. The resource management unit 1202 manages the usage status of resources. A read / write control unit 1203 controls reading and writing from the storage device 1204. Data is stored in the storage device 1204. The declared value management unit 1205 checks the parameters of the read and write messages. Similar to the first embodiment described above, the file system shown in FIG. 12 also secures resources using the GetResource message. The resource management unit 1202 manages the accessible amount to the storage device 1204 and the current usage amount.

While the present invention has been described with respect to several embodiments, it is needless to say that the present invention is not limited to the above embodiments. For example, the following can be said.
(1) The name of the message is an example, and is not limited to the name used in each embodiment. Also, the issue order is not specified.
(2) The parameter of each message is an example of a related portion, and is not limited to this. For example, when the amount of bandwidth to be secured by the GetResource message is determined in advance, T may be stored in the system without being given as a parameter.
(3) The storage device 108 may be physically divided into a plurality.
(4) In the resource management unit 202, the maximum value and the usage amount of the processing capacity are the read data amount per unit time. However, the present invention is not limited to this method. A method of setting the amount of write data per unit time and a method of determining a maximum value for each reading and writing are also conceivable.

  As described above, in the first to third embodiments, when a GetResource message is received, it is possible to guarantee the delay quality for file reading and writing by securing the bandwidth inside the file system. It becomes. By realizing reading or writing of a plurality of files using the reserved bandwidth, it is not necessary to secure a useless bandwidth, and resources inside the file system can be used effectively.

  In addition, at the time of writing, it is necessary to secure a bandwidth obtained by multiplying the required bandwidth by a fixed value, so that it is not necessary to divide the processing capacity of the file system for writing and for reading in advance. Accordingly, the ratio can be easily changed, and resources can be used efficiently.

  Furthermore, by determining whether it is possible to secure the larger bandwidth among the bandwidth that satisfies the delay time declared by the get resource message and the bandwidth that should be secured according to the request of the get resource message, the bandwidth is allocated. It becomes possible to support multiple delay qualities.

  Also, by assigning an MS arrival slot to a block read request, the number of blocks input from the block storage unit to the switching apparatus can be smoothed in time. As a result, it is possible to avoid blocks that are read from a plurality of block storage units and input to the switching apparatus from colliding on the input side to the I / O unit and being discarded inside the switching apparatus. In addition, the fluctuation due to the traffic time of the block can be reduced, and the exchange apparatus can be used efficiently.

  In addition, reservation of MS arrival slots is performed in the order from the largest reserved band to the smallest band, thereby guaranteeing the delay quality of all bands and keeping the maximum number of registrations in each slot constant. % Can be used. At this time, there is no time variation in the number of requested transmissions, and the exchange device can be used most efficiently.

  Furthermore, by assigning an unused MS arrival slot to reading data from a user who has not acquired a resource, it is possible to further improve the use efficiency of the file system.

  For example, in a file system that handles large-scale data that requires continuity, such as a video server, normally, a plurality of storage devices are provided, and data of one file is distributed and stored in each storage device. ing. Under such circumstances, when a plurality of storage devices having different bandwidths and storage capacities (hereinafter simply referred to as capacities) are provided for one file system, control of input / output bandwidths becomes a problem. Conventionally, since control was not performed in consideration of the difference in capacity (bandwidth, capacity) of each storage device, the capacity of each storage device could not be utilized efficiently. Therefore, a disk management device that can easily control a plurality of storage devices having different capacities and input / output bandwidths, and that can optimize the use efficiency of the storage devices in accordance with the required input / output bandwidths, and The file system used will be described.

(Fourth embodiment)
FIG. 13 is a block diagram showing a configuration of a disk management apparatus according to the fourth embodiment of the present invention. In FIG. 13, the present disk management device 1300 includes a disk bandwidth management unit 1301, a disk capacity management unit 1302, a distribution ratio determination unit 1303, a bandwidth calculation unit 1304, a capacity calculation unit 1305, and an all disk bandwidth control unit 1306. And a total disk capacity control unit 1307.

  The disk bandwidth management unit 1301 manages the input / output bandwidth of each storage device provided in the file system. The disk capacity management unit 1302 manages the capacity of each storage device provided in the file system. The distribution ratio determining unit 1303 determines a ratio for distributing data to each storage device in accordance with an algorithm defined by the file system. The bandwidth calculation unit 1304 obtains the lowest value among the values obtained by multiplying the bandwidth of each storage device managed by the disk bandwidth management unit 130 by the reciprocal of the distribution ratio determined by the distribution ratio determination unit 1303. The bandwidth is calculated when all the storage devices are collected. The capacity calculation unit 1305 obtains the smallest value among the values obtained by multiplying the capacity of each storage device managed by the disk capacity management unit 1302 by the reciprocal of the distribution ratio determined by the distribution ratio determination unit 1303. Calculated as the capacity when all storage devices are combined. The total disk bandwidth control unit 1306 performs bandwidth control for all storage devices based on the input / output bandwidth determined by the distribution ratio determination unit 1303 and the bandwidth calculation unit 1304. The total disk capacity control unit 1307 controls the capacity of all storage devices based on the capacity determined by the distribution ratio determination unit 1303 and the capacity calculation unit 1305.

  The operation of the disk management apparatus 1300 configured as described above will be described below. First, an example of the bandwidth setting algorithm executed by the distribution ratio determining unit 1303 will be described.

  FIG. 14 shows a bandwidth priority layout that determines the ratio of distributing data in proportion to the bandwidth of the storage device. For example, when there are three storage devices having different capacities and bandwidths, when data is distributed to each storage device in proportion to the bandwidth, the distribution ratio to each storage device is as shown in FIG. It becomes a ratio value. A value obtained by multiplying the reciprocal of the distribution ratio by the bandwidth of each storage device is defined as an effective bandwidth, and a value obtained by multiplying the reciprocal of the distribution ratio by the capacity of each storage device is defined as an effective capacity. In addition, the bandwidth when all storage devices are combined has the smallest value in the effective bandwidth of each storage device, and the capacity when all storage devices are combined has the smallest value among the effective capacities of each storage device. It is defined as a thing. Then, when all the storage devices are combined, it can be regarded as equivalent to a storage device having a bandwidth of 6 Mbps and a capacity of 6 GB.

  FIG. 15 shows a capacity priority layout that determines the ratio of distributing data in proportion to the capacity of the storage device. For example, if there are three storage devices with different capacities and bandwidths, and the distribution of data to each storage device is performed in proportion to the capacity, the distribution ratio to each storage device is as shown in FIG. It becomes a ratio value. A value obtained by multiplying the reciprocal of the distribution ratio by the bandwidth of each storage device is defined as an effective bandwidth, and a value obtained by multiplying the reciprocal of the distribution ratio by the capacity of each storage device is defined as an effective capacity. In addition, the bandwidth when all storage devices are combined has the smallest value in the effective bandwidth of each storage device, and the capacity when all storage devices are combined has the smallest value among the effective capacities of each storage device. It is defined as a thing. Then, when all the storage devices are combined, it can be regarded as equivalent to a storage device having a bandwidth of 4 Mbps and a capacity of 8 GB.

  FIG. 16 shows a bandwidth setting layout that determines the ratio of distributing data by mixing the bandwidth priority layout and the capacity priority layout. For example, when there are three storage devices having different capacities and bandwidths, distribution to each storage device is executed when the bandwidth priority layout ratio is α and the capacity priority layout ratio is (1−α). The distribution ratio to the apparatus is the distribution ratio value shown in FIG. A value obtained by multiplying the reciprocal of the distribution ratio by the bandwidth of each storage device is defined as an effective bandwidth, and a value obtained by multiplying the reciprocal of the distribution ratio by the capacity of each storage device is defined as an effective capacity. Also, the total bandwidth of all storage devices is the smallest in the effective bandwidth of each storage device, and the total bandwidth of all storage devices is the smallest in the effective capacity of each disk. It is defined as At this time, when it is instructed that the effective bandwidth of all the storage devices exceeds 5 Mbps, α needs to be 3/5 or more. Now, if α is set to 3/5, all the storage devices can be regarded as equivalent to a storage device having a bandwidth of 5 Mbps and a capacity of 20/3 GB.

  Next, the overall operation of the disk management apparatus shown in FIG. 13 will be described. The bandwidth of each storage device connected to the file system is set in advance in the disk bandwidth management unit 1301. In addition, the capacity of each storage device connected to the file system is set in the disk capacity management unit 1302 in advance. When the bandwidth required for the file system is instructed, the distribution ratio determination unit 1303 is set in the disk bandwidth management unit 1302 and the bandwidth information of each storage device set in the disk bandwidth management unit 1301. Based on the information on the capacity of each storage device, the bandwidth calculation unit 1304 and the capacity calculation unit 1305 calculate the bandwidth and capacity of the entire storage device in accordance with the bandwidth setting algorithm. When a data input / output request is actually generated, the total disk bandwidth control unit 1306 controls the bandwidth of the input / output of each storage device according to the bandwidth of the entire storage device calculated by the bandwidth calculation unit 1304. The total disk capacity control unit 1307 controls input / output capacity of each storage device according to the capacity of the entire storage device calculated by the capacity calculation unit 1305.

Note that it is also possible to replace the plurality of storage devices handled individually in the fourth embodiment with what is logically regarded as one storage device.
In the fourth embodiment, all the storage devices are processed together. However, the objects to be collected may be limited to some storage devices instead of all the storage devices.

  As described above, when a combination of storage devices having different input / output bandwidths and capacities is used, management can be performed by managing the bandwidth and capacity of a plurality of storage devices collectively without being conscious of the individual storage devices. In addition, the usage efficiency can be optimized according to the required bandwidth.

(Fifth embodiment)
FIG. 17 is a block diagram showing the configuration of a file system using the disk management apparatus shown in FIG. 17, the file system includes storage devices 1708a and 1708b, a file attribute management unit 1709, a bandwidth management unit 1710, a file open processing unit 1711, a bandwidth acquisition unit 1712, and an input / output control unit 1713. ing.

  Data of each file is distributed and stored in storage devices 1708a and 1708b including storage media such as a hard disk, CD-ROM, MOD, DVD, and semiconductor memory. A file attribute management unit 1709 manages bandwidth characteristics as file attributes. The bandwidth management unit 1710 includes the functions of the disk management device 1300 shown in FIG. 13 as a part thereof, and manages the input / output bandwidth of the file system. The file open processing unit 1711 performs data file open processing. The band acquisition unit 1712 acquires a band used for data input / output. The input / output control unit 1713 combines the plurality of files opened by the file open processing unit 1711 with the band acquired by the band acquisition unit 1712 and selects and inputs arbitrary file data from the plurality of files.

  Actually, in addition to the above, various components such as a function for managing file attributes other than the band characteristics and a function for transmitting and receiving input / output data are required. However, the description is not directly related to the present invention. Is omitted.

The operation of the file system of the fifth embodiment configured as described above will be described below.
First, the disk management device 1300 calculates the distribution ratio of file data to each storage device according to the bandwidth setting algorithm described above. This distribution ratio is described in the layout profile information, and is stored in the file attribute management unit 1709 as the bandwidth characteristic of the file attribute.

  Assume that a first file in which video data is recorded and a second file in which audio data is recorded exist. The bandwidth characteristic of the first file is that the bandwidth required for input / output is 6 Mbps, and the layout profile is evenly distributed to the storage devices 1708a and 1708b. Regarding the bandwidth characteristics of the second file, the bandwidth required for input / output is 0.5 Mbps, and the layout profile is evenly distributed to the storage devices 1708a and 1708b.

  At this time, the data of the first and second files are equally distributed and stored in the storage devices 1708a and 1708b. A file attribute management unit 1709 manages the bandwidth characteristics of the first and second files.

  When the file open processing unit 1711 receives a file open request command for the first file from the user, the file open processing unit 1711 performs a predetermined open process and acquires the corresponding band characteristics from the file attribute management unit 1709. In addition, when the file open processing unit 1711 receives a file open request command for the second file from the user, the file open processing unit 1711 performs a predetermined open process and acquires the corresponding band characteristic from the file attribute management unit 1709.

  Next, when receiving the bandwidth acquisition request command from the user, the bandwidth acquisition unit 1712 determines the bandwidth necessary for simultaneously reading the first and second files according to the bandwidth characteristics acquired by the file open processing unit 1711. Obtained from part 1710.

  Next, when receiving a read request command from the user, the input / output control unit 1713 divides the band acquired by the band acquisition unit 1712 into the first and second files opened by the file open processing unit 1711 and individually The video data of the first file and the audio data of the second file are simultaneously read from the storage devices 1708a and 1708b while synchronizing. Thus, stream data in which video and audio are synchronized is created.

  In the fifth embodiment, the data of each file is evenly distributed to the storage devices 1708a and 1708b. However, it can be distributed unevenly according to the bandwidth and capacity of the storage device.

  Further, in the fifth embodiment, data of two files having different band characteristics are simultaneously read using a band obtained by adding the band characteristics of the two files. Alternatively, the band may be read alternately using the band determined by the band characteristics of one file (the largest band). Thus, by alternately reading out data of a plurality of files using one band, it is possible to easily create an application such as AV synchronization.

  Incidentally, as an example of a file system, in a video information providing system that provides video information to a plurality of clients, a buffer for temporarily storing transmitted information is generally provided in each client. Each client transmits a service request to a server in the file system at a period determined by the capacity of each buffer and the bit rate of the received video, and the server receives a predetermined amount of video information according to the service request. Is sent to each client. However, since the server has an upper limit on the amount of information that can be transmitted at once, if service requests compete, it is necessary to schedule service requests from each client so that video discontinuity does not occur. is there.

  Therefore, a server in a conventional video information providing system is provided with a scheduling device for scheduling service requests that are required to be processed periodically. As a conventional scheduling apparatus, the one described in Japanese Patent Publication No. 7-104792 is known. In this apparatus, the schedule control program is provided with a priority processing queue for each task and for each priority so that a process with a higher priority can be executed first. is there.

  By using the above-described conventional scheduling device, a service request from a client having a high priority, i.e., a client having a shorter cycle for transmitting a service request, is executed preferentially. Therefore, the frequency of video discontinuity can be reduced.

  However, since the conventional scheduling apparatus always executes a service request from a higher-priority client, which is transmitted in a shorter cycle, with higher priority, a lower-order client is not executed for a long time. There is a case.

  Moreover, the time until the service request from the lower client is executed varies depending on the number of service requests from other clients. For this reason, it is impossible to prevent the discontinuity of the video from occurring by transmitting the service request at a period taking into account the delay time on the lower client side.

  Further, when the time average of the number of service requests transmitted from all clients exceeds the processing capability of the server, the number of service requests that cannot be processed increases as time passes. In this case, as a matter of course, it is impossible to prevent the discontinuity of the video from occurring in any order.

  Further, in a video information providing system in which a plurality of servers provide video information to clients via a common transmission path, there is an upper limit on the amount of information that can be transmitted at one time on the transmission path. Need to schedule service requests. Therefore, when scheduling is performed using the conventional scheduling device, the same problem as described above occurs.

  Therefore, in the following, when continuous information such as video information is divided and provided from a server to a plurality of clients, a service request from each client is scheduled so that discontinuity of information does not occur. A possible scheduling apparatus will be described.

  Also, a description will be given of a scheduling apparatus capable of scheduling service requests to each server so that discontinuity of information does not occur when continuous information is divided and provided from a plurality of servers to clients.

(Sixth embodiment)
Hereinafter, a sixth embodiment of the present invention will be described with reference to the drawings.
FIG. 18 is a block diagram showing a configuration of a scheduling apparatus according to the sixth embodiment of the present invention. The apparatus in FIG. 18 includes a registration instruction unit 1811, an execution right generation cycle table 1812, an execution right generation unit 1813, a process waiting queue 1814, and an execution instruction unit 1815. The execution instruction unit 1815 includes an execution right queue 1816.

  FIG. 19 is a block diagram showing a configuration of a video information providing system using the scheduling apparatus of FIG. The system in FIG. 19 includes a video information providing device 1920 as an example of a file system and a plurality of clients (1 to n) 1921. The video information providing device 1920 includes a scheduling device 1810 and a server 1922.

  Each of the clients (1 to n) 1921 in FIG. 19 periodically transmits a read request to the server 1922 of the video information providing apparatus 1920. The client (1 to n) 1921 is provided with a buffer and a monitor, and the client (1 to n) 1921 temporarily stores the provided video information in the buffer and reads the stored information while displaying the video on the monitor. indicate. The scheduling apparatus 1810 schedules read requests so that video information is periodically provided from the server 1922 to the clients (1 to n) 1921, respectively. The server 1922 transmits video information by a predetermined amount to the clients (1 to n) 1921 in accordance with an instruction from the scheduling device 1810.

  18 receives the report from the clients (1 to n) 1921 in FIG. 19 and the period for generating the execution right for each client (1 to n) 1921 is displayed in the execution right generation period table. Instruct 1812 to register. The execution right generation cycle table 1812 registers a cycle in which an execution right should be generated. The execution right generation unit 1813 is activated at a predetermined cycle (hereinafter, “Tslt”), and generates an execution right for each client (1 to n) 1921 based on the cycle registered in the execution right generation cycle table 1812. The execution right generation unit 1813 also assigns an evaluation index (hereinafter referred to as Nsft) indicating the priority to the generated execution right. The processing queue 1814 queues read requests from the clients (1 to n) 1921. The execution instruction unit 1815 is activated at the cycle Tslt, and acquires a read request from the client 1921 corresponding to the execution right generated by the execution right generation unit 1813 among the read requests queued in the processing queue 1814. The server 1922 is instructed to execute the read request.

  FIG. 20 is a sequence diagram showing how signals are transmitted and received between the video information providing apparatus 1920 and the clients (1 to n) 1921 in FIG. The resource acquisition request 31 in FIG. 20 includes a client (1 to n) in order to notify information (hereinafter referred to as a client ID) for identifying the transmission source client 1921, a video information name desired to be read, and a reading position. ) 1921 is a signal transmitted to the video information providing apparatus 1920. The resource acquisition request 31 further includes a cycle for transmitting a read request desired by each of the clients (1 to n) 1921, that is, a cycle for generating the execution right (hereinafter referred to as Treq). The resource acquisition response 32 is a signal transmitted from the video information providing apparatus 1920 to the clients (1 to n) 1921 in order to notify whether or not video information can be provided with a desired Treq.

  The read request 33 is a signal transmitted from the client (1 to n) 1921 to the video information providing apparatus 1920 when desiring to read. The read response 34 is video information that the video information providing apparatus 1920 transmits to the clients (1 to n) 1921 in response to the read request 33. The resource release request 35 is a signal transmitted from the client (1 to n) 1921 to the video information providing apparatus 1920 in order to notify that reading of desired video information is completed, and includes a client ID. The resource release response 36 is a signal transmitted from the video information providing apparatus 1920 to the clients (1 to n) 1921 in order to notify that the provision of the video information is stopped in response to the resource release request 35.

FIG. 21 is a diagram showing the registration contents of the execution right generation cycle table 1812 of FIG. As illustrated in FIG. 21, the execution right generation cycle table 1812 registers the client ID, Treq, and execution right generation cycle offset (hereinafter, Tofs) for each client (1 to n) 1921.
FIG. 22 is a diagram showing the queuing contents of the process waiting queue 1814 of FIG. As illustrated in FIG. 22, the processing queue 1814 queues a read request from the clients (1 to n) 1921 for each client (1 to n) 1921.

FIG. 23 is a diagram showing the queuing contents of the execution right queue 1816 of FIG. As shown in FIG. 23, the execution right queue 1816 queues the generated execution right for each assigned Nsft.
FIG. 24 is a flowchart for explaining an operation in which the execution right generation unit 1813 in FIG. 18 generates an execution right.

  FIG. 25 is a flowchart for explaining an operation in which the execution instructing unit 1815 in FIG. 18 instructs the server 1922 in FIG. 19 to execute the read request.

Hereinafter, an operation in which the scheduling apparatus 1810 in FIG. 19 schedules a read request so that video information is periodically provided from the server 1922 to the plurality of clients (1 to n) 1921 will be described with reference to FIG. Description will be made with reference to .about.25.
First, an operation in which the registration instruction unit 1811 in FIG. 18 instructs the execution right generation cycle table 1812 to register the client ID, Treq, and Tofs will be described.

  Now, it is assumed that the client IDs, Treq, and Tofs of a plurality of clients 1921 are already registered in the execution right generation cycle table 1812. At this time, another client 1921 transmits its own resource ID and Treq to the video information providing device 1920 by transmitting the resource acquisition request 31 of FIG.

  The registration instructing unit 1811 recognizes the information provision capability of the server 1922, that is, the maximum number of read requests (hereinafter referred to as Nmax) that can be executed by the server 1922 within a time equal to the period Tslt, and has already been registered. In addition, when the declared Treq is registered, the server 1922 can provide information on an average over time by comparing the average number of execution rights generated per Tslt with Nmax. It is determined whether or not.

  If it is determined that information can be provided, the registration instruction unit 1811 instructs the execution right generation cycle table 1812 to register the reported client ID and Treq, and registers “0” as the initial value of Tofs. Let Then, the registration instruction unit 1811 notifies the transmission source client 1921 of the resource acquisition request 31 that the video information can be provided by the declared Treq by transmitting the resource acquisition response 32.

  If it is determined that the information cannot be provided, the registration instruction unit 1811 notifies the transmission source client 1921 that the information cannot be provided by the declared Treq by transmitting the resource acquisition response 32.

  Next, an operation in which the execution right generation unit 1813 in FIG. 18 generates an execution right based on Treq and Tofs registered in the execution right generation cycle table 1812 will be described with reference to FIG. In the execution right generation cycle table 1812, it is assumed that the client IDs, Treq, and Tofs of a plurality of clients 1921 are registered as described above.

The execution right generation unit 1813 first extracts a set of client ID, Treq, and Tofs registered in the execution right generation cycle table 1812 (step S1). Next, the execution right generation unit 1813 compares the extracted Tofs with the cycle Tslt of its own operation (step S2).
If Tofs ≧ Tslt, the execution right generation unit 1813 proceeds to step S9.

If Tofs <Tslt, the execution right generation unit 1813 determines whether or not a read request from the client 1921 corresponding to the read client ID is queued in the processing queue 1814 (step S3). .
When queued, the execution right generation unit 1813 generates an execution right and gives information indicating that the execution right has been assigned to the read request. On the other hand, a new Tofs is obtained by the following equation (6). Is calculated (step S4).

Tofs = Tofs + Treq (6)
If it is not queued, the execution right generation unit 1813 proceeds to step S9.

Next, the execution right generation unit 1813 compares the new Tofs obtained by calculation with Tslt (step S5).
If Tofs <Tslt, the execution right generation unit 1813 gives Nsft = 0 as an evaluation index to the generated execution right (step S6). When the execution instruction unit 1815 instructs the execution right queue 1816 to cause the generated execution right to be queued at the tail of the queue having the same value of Nsft (step S7), the execution right generation unit 1813 Return to S2.

When Tofs ≧ Tslt, the execution right generation unit 1813 assigns Nsft calculated by the following expression (7) as an evaluation index to the generated execution right (step S8), and proceeds to step S7.
Nsft = f (Treq / Tslt) (7)
However, in the above formula (7), f (x) is a minimum integer equal to or greater than x.

If Tofs ≧ Tslt in step S2, the execution right generation unit 1813 calculates new Tofs by the following equation (8) (step S9).
Tofs = Tofs−Tslt (8)

Next, the execution right generation unit 1813 determines whether or not the new Tofs obtained by the calculation is Tofs ≧ 0 (step S10).
When Tofs ≧ 0, the execution right generation unit 1813 instructs the execution right generation cycle table 1812 to update Tofs to new Tofs obtained by calculation (step S11).

  If Tofs <0, the execution right generation unit 1813 updates Tofs to 0 (step S12).

Next, the execution right generation unit 1813 determines whether or not all client IDs, Treq, and Tofs registered in the execution right generation cycle table 1812 have been extracted (step S13).
If it is determined that all have been extracted, the execution right generation unit 1813 stops the operation, and if it is determined that there is something to be extracted, the process returns to step S1.

Next, an operation in which the execution instruction unit 1815 in FIG. 18 instructs the server 1922 to execute a queued read request based on the execution right generated as described above will be described with reference to FIG. The description will be given with reference.
The execution instructing unit 1815 counts the number of times the server 1922 has been instructed to execute, and stores a count value C obtained by counting. The execution instructing unit 1815 first initializes the count value C (step S21).

  Next, the execution instructing unit 1815 is the head of the execution right queued in the queue having the smallest Nsft value among the execution rights queued in the execution right queue 1816, that is, the assigned Nsft. Is the smallest and the longest elapsed time since it was generated (step S22), and then the count value C is incremented (step S23).

  Next, the execution instructing unit 1815 refers to the client ID assigned to the extracted execution right, extracts the read request from the client 1921 corresponding to the execution right from the processing queue 1814, and executes it to the server 1922. (Step S24). Then, the execution instruction unit 1815 compares the incremented count value C with Nmax (step S25).

  If C <Nmax, the execution instruction unit 1815 returns to step S22. If C ≧ Nmax, the execution instruction unit 1815 instructs the execution right queue 1816 to decrement the Nsft value of the remaining execution right. (Step S26), and then the process ends.

  The server 1922 that has received the execution instruction as described above transmits the specified video information to the client 1921 that is the source of the read request after a certain time has elapsed since the instruction was received. When the reception of the necessary video information is completed, the client 1921 notifies the video information providing apparatus 1920 that the reception has been completed by transmitting a resource release request 35 of FIG. The resource release request 35 is assigned with the client ID of the transmission source client 1921, and the registration instruction unit 1811 that has received the resource release request 35 sends the assigned client ID and its client ID from the execution right generation cycle table 1812. Delete Treq and Tofs corresponding to the client ID.

  As described above, according to the present embodiment, when a read request is scheduled so that video information is periodically provided from the server 1922 to a plurality of clients (1 to n) 1921, a processing queue 1814 queues a read request from each client (1 to n) 1921. The execution right generation cycle table 1812 registers a cycle Treq for generating the execution right for each client (1 to n) 1921. The execution right generation unit 1813 is based on the cycle registered in the execution right generation cycle table 1812. To generate an execution right. The execution instruction unit 1815 acquires a read request from the client 1921 corresponding to the execution right generated by the execution right generation unit 1813 among the read requests queued in the processing queue 1814 and executes the read request on the server 1922. Instruct.

  As described above, while queuing the read request from each client (1 to n) 1921, the cycle for generating the execution right for each client (1 to n) 1921 is registered, and based on the cycle. To generate an execution right. Then, the server 1922 is instructed to execute a request from the client 1921 corresponding to the generated execution right among the queued read requests. As a result, the time until a read request from one client 1921 is executed is not affected by the status of transmission of the read request from another client 1921, and therefore the transmitted read request is executed. It is possible to prevent the time until the time is indefinite.

  Further, the execution right generation unit 1813 is activated at a cycle Tslt. With reference to the execution right generation cycle table 1812, the time Tofs from the current activation time to the time when the next execution right is generated is expressed as Tofs <Tslt. The execution right of the client 1921 is generated. Further, Treq is added to the Tofs of the client 1921. When Tofs obtained by the addition is Tofs ≧ Tslt, a new Tofs is calculated for the Tofs by the formula Tofs = Tofs−Tslt. The Tofs obtained in this manner are registered in the execution right generation cycle table 1812 by instructing them.

  When the Tofs obtained by the addition is Tofs <Tslt, the execution right generation unit 1813 further generates an execution right and repeats the operation of adding Treq to Tofs and comparing it with Tslt. Then, when Tofs obtained by the addition becomes Tofs ≧ Tslt, new Tofs is calculated with respect to the Tofs by the expression Tofs = Tofs−Tslt, and the Tofs obtained by the calculation is changed to the execution right. The generation period table 1812 is instructed to be registered.

  For a client 1921 in which the time Tofs from the current activation time to the time for generating the next execution right is Tofs ≧ Tslt, or the read request is not queued in the processing queue 1814 The execution right is not generated, new Tofs are calculated for the Tofs by the formula Tofs = Tofs-Tslt, and the Tofs obtained by calculation are instructed and registered in the execution right generation period table 1812.

  If the calculated Tofs is Tofs <0, the execution right generation unit 1813 instructs Tof = 0 to be registered in the execution right generation cycle table 1812.

  As described above, the execution right generation unit 1813 is activated at the cycle Tslt, and when activated, based on Treq registered in the execution right generation cycle table 1812 and Tslt, that is, the current time at which the execution right generation unit 1813 was activated. To determine the client 1921 that should generate the execution right by calculating the time until the time when the next execution right is generated and comparing the Tofs obtained by calculation with Tslt and generating the execution right Let Furthermore, the execution right generation unit 1813 adds the Treq to Tofs and repeats the operation of comparing with Tslt, thereby determining the client 1921 that should generate a plurality of execution rights and generating the necessary number of execution rights. Let The execution right generation unit 1813 performs these operations collectively at the time of activation, so that the execution right generation unit 1813 generates the execution right in the cycle Treq registered in the execution right generation cycle table 1812. Processing operations can be reduced.

  Further, by not generating the execution right for the client 1921 for which the read request is not queued, it is possible to prevent the video information from being provided to the client 1921 that has not transmitted the read request.

  In addition, when the new Tofs obtained by calculation is Tofs <0, by registering Tofs = 0, the Tofs of the client 1921 in which the read request is not queued is repeatedly subtracted from Tslt. It becomes a small value, and when the read request is queued, it is possible to prevent a large number of execution rights of the client 1921 from being generated.

  Further, the execution right generation unit 1813 gives the evaluation index Nsft calculated by the above equation (7) to the execution right generated as described above. The execution instruction unit 1815 includes an execution right queue 1816 for queuing the execution right generated by the execution right generation unit 1813, and instructs the generated execution right to be generated for each evaluation index. Queue in the order in which they were made. Then, the execution instructing unit 1815 selects the execution right that has been given the smaller evaluation index Nsft from among the execution rights that are queued, and the client corresponding to the execution right is selected. The server 1922 is instructed to execute a read request from 1921.

  The execution instructing unit 1815 recognizes the maximum number Nmax of read requests that can be executed by the server 1922 within a time equal to the cycle Tslt, and stops the execution instruction when the execution is instructed by the number equal to Nmax. And the execution right queue 1816 is instructed to decrement the remaining execution right evaluation index Nsft.

  In this way, the execution right generation unit 1813 assigns the evaluation index indicating the priority calculated as a function of the cycle in which the execution right should be generated to the generated execution right, and the execution instruction unit 1815 has been given Based on the evaluation index, an execution right to be prioritized is selected and execution is instructed. Then, when the execution instruction unit 1815 instructs execution of Nmax read requests, the execution instruction unit 1815 temporarily stops the execution instruction, and further decrements the evaluation index for the remaining execution right, Make it higher.

  As a result, the execution instruction unit 1815 can preferentially select the execution right of the client 1921 having a shorter time from the current activation time to the time when the next execution right is generated. The number of delays can be reduced. In addition, it is possible to prevent the server 1922 from executing a number of read requests exceeding the processing capacity, and the remaining execution rights are decremented every Tslt, so that they are generated. The execution right can be selected at the latest by Treq + Tslt. Therefore, the client (1 to n) 1921 can prevent the discontinuity of the video information from occurring by transmitting the read request at a cycle that takes this maximum delay time into consideration.

  However, when an execution right exceeding the processing capability of the server 1922 is generated on an average over time, the number of remaining execution rights increases as time passes. In this case, even if the above processing is performed, it is impossible to prevent the discontinuity of the video from occurring.

  Therefore, when the registration instructing unit 1811 receives a declaration to register Treq, the average number of execution rights generated per Tslt even if the registered Treq is further registered in addition to those already registered. However, only when it is determined that Nmax does not exceed Nmax, the execution right generation cycle table 1812 is instructed to register the declared Treq.

  As a result, since the total number of execution rights generated exceeds the processing capacity of the server 1922 on an average over time, the number of execution rights remaining increases with time, resulting in video discontinuity. Can be prevented.

  Further, the execution instruction unit 1815 is provided with an execution right queue 1816 and is queued for each evaluation index and in the order in which it is generated, so that the execution instruction unit 1815 selects a queue with the smallest evaluation index when selecting an execution right. Since selection may be made in order from the top, the processing operation of the execution instruction unit 1815 can be reduced.

(Seventh embodiment)
The seventh embodiment of the present invention will be described below with reference to the drawings.
FIG. 26 is a block diagram showing a configuration of a scheduling apparatus according to the seventh embodiment of the present invention. The apparatus of FIG. 26 includes a registration instruction unit 2691, an execution right generation cycle table 2692, an execution right generation unit 2693, a processing queue 2694, and an execution instruction unit 2695. The execution instruction unit 2695 includes an execution right queue 2696.

  FIG. 27 is a block diagram showing a configuration of a video information providing system using the scheduling apparatus of FIG. The system in FIG. 27 includes a video information providing apparatus 2700 and a client 2701, and the video information providing apparatus 2700 and the client 2701 are connected by a transmission path 2703. The video information providing apparatus 2700 includes a scheduling apparatus 2690 and a plurality of servers (1 to m) 2702.

  The client 2701 in FIG. 27 periodically transmits read requests to the plurality of servers (1 to m) 2702 of the video information providing apparatus 2700. The client 2701 is provided with a plurality of buffers and monitors, and the client 2701 temporarily stores the provided information in each buffer for each server (1 to m) 2702 and reads the stored information. Display video on each monitor. The scheduling apparatus 2690 schedules read requests so that video information is periodically provided from the server (1 to m) 2702 to the client 2701, respectively. Each of the servers (1 to m) 2702 transmits video information by a predetermined amount to the client 2701 in accordance with an instruction from the scheduling device 2690. It is assumed that the total information providing capability of the servers (1 to m) 2702 included in the video information providing apparatus 2700 exceeds the transmission capability of the transmission path 2703.

  26 receives the report from the client 2701 in FIG. 27, and instructs the execution right generation period table 2692 about the period for generating the execution right for each server (1 to m) 2702. Let me register. The execution right generation cycle table 2692 registers a cycle in which an execution right should be generated. The execution right generation unit 2693 is activated at a predetermined cycle (hereinafter, Tslt), and generates an execution right for each server (1 to m) 2702 based on the cycle registered in the execution right generation cycle table 2692. The execution right generation unit 2693 also assigns an evaluation index (hereinafter referred to as Nsft) indicating priority to the generated execution right. The processing queue 2694 queues a read request to the server (1 to m) 2702. The execution instruction unit 2695 is activated at the cycle Tslt, and acquires a read request to the server 2702 corresponding to the execution right generated by the execution right generation unit 2893 among the read requests queued in the processing queue 2694. The corresponding server 2702 is instructed to execute the read request.

  In addition, the same signal as described with reference to FIG. 20 in the sixth embodiment is transmitted and received between the video information providing apparatus 2700 and the client 2701 in FIG. However, the resource acquisition request 31 includes a server ID instead of the client ID.

  FIG. 28 is a diagram showing the registration contents of the execution right generation cycle table 2692 of FIG. As shown in FIG. 28, the execution right generation cycle table 2692 registers the server ID, Titv, and execution right generation cycle offset (hereinafter, Tofs) for each server (1 to m) 2702. Note that Treq in the table in FIG. 21 is a value declared from the clients (1 to n) 1921 in FIG. 19, but Titv in FIG. 28 is a predetermined value based on Treq reported from the client 2701 in FIG. It is a value calculated by the method.

  FIG. 29 is a diagram showing the queuing contents of the process waiting queue 2694 in FIG. As shown in FIG. 29, the processing queue 2694 queues a read request to the server (1 to m) 2702 for each server (1 to m) 2702.

  FIG. 30 is a diagram showing the queuing contents of the execution right queue 2696 in FIG. As shown in FIG. 30, the execution right queue 2696 queues the generated execution right for each assigned Nsft.

In the following, the scheduling apparatus 2690 in FIG. 27 schedules read requests so that video information is periodically provided from the plurality of servers (1 to m) 2702 to the client 2701 via the transmission path 2703, respectively. The operation of performing will be described with reference to FIGS.
First, an operation in which the registration instruction unit 2691 in FIG. 26 instructs the execution right generation cycle table 2692 to register the server ID, Titv, and Tofs will be described.

  Now, it is assumed that the server IDs, Titv, and Tofs of a plurality of servers 2702 are already registered in the execution right generation cycle table 2692. At this time, the client 2701 reports the server ID and Treq of the desired server 2702 to the video information providing apparatus 2700 by transmitting the resource acquisition request 31 of FIG.

The registration instruction unit 2691 recognizes the information providing capability of the server (1 to m) 2702 and the transmission capability of the transmission path 2703, and determines the following (a) and (b).
(A) Whether the information providing capability of the designated server 2702 can provide information when the already registered Titv is updated to a new Titv including the declared Treq.
(B) The transmission capability of the transmission line 2703 can transmit information provided from the server (1 to m) 2702 when the already registered Titv is updated to a new Titv including the declared Treq. Is it?

When it is determined that both of the above capabilities (a) and (b) are present, the registration instruction unit 2691 instructs the execution right generation cycle table 2692 to register the Titv of the designated server 2702 that has already been registered. Is updated to a new Titv calculated by the following equation (9), and the client 2701 is notified that the request is accepted by transmitting the resource acquisition response 32 of FIG.
Titv = 1 / (1 / Titv + 1 / Treq) (9)

Also, the registration instruction unit 2691 instructs the execution right generation cycle table 2692 to update Tofs to new Tofs calculated by the following equation (10).
Tofs = Tofs + (Titv (new) −Titv (old)) (10)
However, in the above formula (10), Titv (new) means updated Titv, and Titv (old) means Titv before being updated.

  If it is determined that there is not at least one capability, the registration instruction unit 2691 notifies the client 2701 that the request is not accepted by transmitting a resource acquisition response 32.

  If it is determined that both of the above capabilities (a) and (b) are present, and the server ID of the designated server 2702 is not registered in the execution right generation cycle table 2692, the registration instruction unit 2691 Instructs the execution right generation cycle table 2692 to newly register the server ID, Titv, and Tofs. In this case, the declared Treq is registered as Titv, and “0” is registered as Tofs.

  The operation of generating the execution right based on the Titv and Tofs registered in the execution right generation cycle table 2692 by the execution right generation unit 2693 is the same as that described with reference to FIG. 24 in the sixth embodiment. . However, the execution right generation cycle table 2692 is different in that the server ID, Titv, and Tofs of a plurality of servers (1 to m) 2702 are registered as described above.

Also, in step S4 in FIG. 24, a new Tofs is calculated by the following equation (11) instead of the equation (6).
Tofs = Tofs + Titv (11)

  Similarly, in step S8, Nsft is calculated by the following equation (12) instead of equation (7).

Nsft = f (Titv / Tslt) (12)
However, in the above formula (12), f (x) is a minimum integer equal to or greater than x.

  Next, the execution instruction unit 2695 selects an execution right based on the assigned evaluation index, and reads out to the server 2702 corresponding to the selected execution right among the read requests queued in the processing queue 2694. The request is acquired and the corresponding server 2702 is instructed to execute. The operation of the execution instruction unit 2695 at that time is the same as that described with reference to FIG. 25 in the sixth embodiment. However, the count value C is the total number of times that the execution instructing unit 2695 has instructed all servers (1 to m) 2702 to execute, and Nmax is the maximum information that the transmission line 2703 can transmit within the time equal to the period Tslt. The number of read requests corresponding to the quantity.

  When the video information is provided as described above, and the provision is completed, the client 2701 transmits the resource release request 35 of FIG. 20 in the same way as the client 1921 performs in the sixth embodiment. The video information providing apparatus 2700 is notified that the reception of the video information to be completed is completed. However, the server ID and Treq are added to the resource release request 35 transmitted by the client 2701, and the registration instruction unit 2691 in FIG. 26 instructs the execution right generation period table 2692 to specify the registered Titv. , Updated to a new Titv calculated by the following equation (13).

Titv = 1 / (1 / Titv-1 / Treq) (13)
Also, the registration instruction unit 2691 instructs the execution right generation cycle table 2692 to update Tofs to new Tofs calculated by the above equation (10).

  As described above, according to the present embodiment, when scheduling read requests so that video information is periodically provided from the plurality of servers (1 to m) 2702 to the client 2701, the sixth The same operation as in the embodiment is performed. Further, when the registration instruction unit 2691 receives a report to update the already registered period Ttv for generating the execution right of a server 2702 to one that generates the execution right in the period Treq, the registration instruction unit 2691 has already registered. Even if the registered Titv is updated to a new Titv including the declared Treq, the transmission capability of the transmission path 2703 can transmit the information provided from all the servers (1 to m) 2702 Only, the execution right generation cycle table 2692 is instructed to update the registered Titv to a new cycle Titv calculated by the above equation (9).

  As a result, when the registered cycle is updated, the average number of execution rights generated per Tslt exceeds the transmission capability of the transmission path, so that the number of execution rights remaining over time is reduced. It can be prevented from increasing.

  In addition, when the registration instruction unit 2691 receives a report to update the already-registered period Titv for generating the execution right of a server 2702 to one that generates the execution right excluding the period Treq, The execution right generation cycle table 2692 is instructed to update the registered Titv to a new cycle Titv calculated by the above equation (13).

  As a result, when a declaration is received to update the already registered cycle for generating the execution right to one that is generated excluding the execution right generated in the cycle Treq, the already registered cycle is changed. , Can be updated with new ones.

  In this embodiment, only one client 2701 having a plurality of buffers and monitors is included. However, a plurality of clients 2701 having one buffer and monitors are included via a common transmission line 2703. You may make it. However, in this case, each client 2701 needs to transmit a read request at a period determined for each client.

  In this embodiment, scheduling is performed due to the limitation of the transmission capability of the transmission line 2703. However, scheduling is limited because of the limitation of the reception capability (buffer capacity, processing speed, etc.) of the client 2701. It may be done.

  In the sixth and seventh embodiments, the execution right generation units 1813 and 2693 and the execution instruction units 1815 and 2695 are both activated at the cycle Tslt. Also good.

  In the sixth embodiment, Nmax is the maximum number of read requests that can be executed by the server 1922 within a time equal to the period Tslt. However, the maximum number of read requests that can be executed by the server 1922 in unit time. It may be. Similarly, in the seventh embodiment, Nmax is the number of read requests corresponding to the amount of information that can be transmitted by the transmission line 2703 within a time equal to the period Tslt. It may be the number of read requests corresponding to a possible amount of information.

  Further, by using other scheduling devices including the scheduling device of FIG. 18 and the scheduling device of FIG. 26, scheduling is performed so that video information is periodically provided from a plurality of servers to a plurality of clients, respectively. It can be carried out. That is, first, the scheduling device in FIG. 18 generates an execution right for each client, selects an execution right of a client from the generated execution rights, and issues a read request corresponding to the execution right in FIG. To the scheduling device. Next, the scheduling device in FIG. 26 generates an execution right for each server, and selects an execution right of a certain server from the generated execution rights, and executes the input read request to the server. Instruct.

  In the sixth and seventh embodiments, the execution right generation units 1813 and 2693 calculate Nsft according to the above formulas (7) and (12), respectively. However, Treq is obtained by dividing Treq by Tslt. The integer part of the obtained value may be Nsft.

  Also, the execution right generation period tables 1812 and 2692 register Tnext indicating the time for generating the next execution right instead of the execution right generation period offset Tofs, and the execution right generation units 1813 and 2693 have the current time Tnext. An execution right may be generated at the time of matching.

  Further, the execution right generation units 1813 and 2693 calculate Nsft, and the execution instruction units 1815 and 2695 select the execution right having a smaller Nsft. However, the execution right generation units 1813 and 2693 calculate the above Tnext, The execution instruction units 1815 and 2695 may select an execution right whose Tnext is closer to the current time.

  Further, when the count value C satisfies C ≧ Nmax, the execution instruction units 1815 and 2695 instruct the execution right queue 1816 and the execution right queue 2696, respectively, to decrement Nsft of the remaining execution right. However, instead of the execution right queues 1816 and 2696, another execution right queue as shown in FIG. 31 is used, and when the count value C is C ≧ Nmax, the execution instruction units 1815 and 2695 The execution right queue may be instructed to shift the pointer. As shown in FIG. 31, the other execution right queue has a ring structure, and the value of Nsft is sequentially changed as the position of the pointer moves.

  As described above, in a file system such as a video-on-demand system in a CATV network or a video editing system in a broadcasting station, a server device that handles data such as video and audio is provided. When receiving a data read request from a user terminal or client, this server device searches an internal storage device, reads data corresponding to the request, and sends the data to the user terminal or client that requested the read. It has a function.

  By the way, in order to guarantee the continuity of the data to be transmitted, the conventional server device has a resource (data per unit time) in the server device that matches the amount of data to be transmitted before transmitting the data to the user terminal. Data and the right to use the communication resource (capability to transmit data per unit time) between the server device and the user terminal that requests data acquisition. Is to be executed.

  However, in the conventional server device as described above, not only data that requires continuity such as moving images and sounds, but also data that does not require continuity such as still images is transmitted before data transmission. Resource usage rights are secured. Therefore, as the amount of data that does not require continuity increases, it becomes difficult to secure data resources that require continuity, and in the worst case, data cannot be transmitted.

  Therefore, in the following, resources can be used efficiently when sending data that does not require continuity, and continuity is required even if the amount of data that does not require continuity increases. A server apparatus that does not affect the securing of resources for sending data will be described.

(Eighth embodiment)
FIG. 32 is a block diagram showing a configuration of a data transmission system using a server device according to the eighth embodiment of the present invention. 32, the data transmission system includes a resource management unit 3201, a data storage unit 3202, an exchange unit 3203, an input / output unit 3204, and a plurality of user terminals 3208. Note that the server device of this embodiment includes a resource management unit 3201, a data storage unit 3202, an exchange unit 3203, and an input / output unit 3204. The server device and each terminal 3208 are connected via a wired or wireless transmission path.

  The resource management unit 3201 manages necessary resources when data is transmitted to each user terminal 3208. The data storage unit 3202 includes a plurality of storage devices (for example, a hard disk drive and an optical disc driver with an autochanger) 3205 that can be randomly accessed, and a plurality of MSFSs 3206 that are connected to each storage device 3205 and control reading of data. The exchange unit 3203 exchanges data between the input / output unit 3204 and the data storage unit 3202. In this embodiment, as an example, an ATM switch is used for the exchange unit 3203. The input / output unit 3204 includes a plurality of I / O units 3207 and controls data input / output between the server device and each user terminal 3208. Each user terminal 3208 is connected to one of the I / O units 3207.

  In the above configuration, a connection is set in a mesh shape by the exchange unit 3203 between each I / O unit 3207 and each MSFS 3206. Note that the bandwidth of each connection is managed by the resource management unit 3201.

  FIG. 33 is a block diagram showing a more detailed configuration of the I / O unit 3207 in FIG. In FIG. 33, the I / O unit 3207 includes a priority input / output control unit 3301 and a non-priority input / output control unit 3302. The non-priority input / output control unit 3302 includes a command queue 3321, a command queue control unit 3322, a command transmission unit 3323, a data buffer unit 3324, a buffer management unit 3325, a data transmission unit 3326, and a command reception unit 3327. including.

  The priority input / output control unit 3301 controls input / output of data requiring continuity (hereinafter referred to as priority data) such as moving image data. On the other hand, the non-priority input / output control unit 3302 controls input / output of data that does not require continuity such as still image data (hereinafter referred to as non-priority data).

  In the non-priority input / output control unit 3302, the command queue 3321 stores a command for requesting reading of non-priority data from the user terminal 3208 for each user terminal. The command queue control unit 3322 controls the extraction of commands from the command queue 3321. The command sending unit 3323 sends the command extracted by the command queue control unit 3322 to the corresponding destination MSFS 3206. The data buffer unit 3324 stores data read from the storage device 3205 by the MSFS 3206. The buffer management unit 3325 counts the data stored in the data buffer unit 3324 (or reserved for storage) for each user terminal that has requested reading. The data sending unit 3326 sends the data stored in the data buffer unit 3324 to the corresponding destination user terminal 3208, and sets the area of the data buffer unit 3324 that has been sent to an empty state. The command reception unit 3327 manages whether or not a command that has arrived from the user terminal 3208 is registered in the command queue 3321.

The operation of the server device configured as described above will be described below.
First, the resource initial setting operation performed when the system is started will be described. At the time of system startup, the resource management unit 3201 has resources of each unit in the data transmission system, that is, a data read band of each MSFS 3206, a communication band between each I / O unit 3207, and each ATM in the switching unit 3203. Switch communication bandwidth, etc.
(1) Resources used to output priority data (hereinafter referred to as priority resources);
(2) It allocates separately to resources used to output non-priority data (hereinafter referred to as non-priority resources). Thereafter, the resource management unit 3201 manages each separated resource individually.

  Note that the data read bandwidth of the MSFS 3206 is within a predetermined delay time when data is read in response to a read request requested by the MSFS 3206, where the storage location of the read data is random and the request interval is random. Is defined as the maximum speed of data that can be read at a predetermined discard rate or less.

  Next, priority data output operation will be described. When each user terminal 3208 requests the server device to output priority data, the user terminal 3208 requests to secure resources of a band necessary for transmitting the priority data. Upon receiving the resource securing request, the priority input / output control unit 3301 requests the resource management unit 3201 to secure the resource by specifying the bandwidth amount. In the resource management unit 3201, the ATM switch communication band, the MSFS 3206 data read band, and the I / O unit 3207 communication band are pre-assigned as priority resources in order to transmit the requested amount of band data. Secure from the amount of bandwidth available.

  If the priority resource is currently being used and there is not enough room for the requested bandwidth, the resource reservation fails and the server device cannot execute the output of the priority data. On the other hand, when the resource is successfully secured, for the data that the user terminal 3208 desires to output, the communication band of the I / O unit 3207, the data read band of the MSFS 3206, and the communication band of the ATM switch of the switching unit 3203 are: Set for each device.

  Upon receiving the resource reservation completion notification from the priority input / output control unit 3301, the user terminal 3208 issues a command for requesting output of priority data. Upon receiving the command, the priority input / output control unit 3301 of the I / O unit 3207 executes data output using the resources reserved for the priority data to be output.

  FIG. 34 is a flowchart showing the operation of the command receiving unit 3327 in FIG. FIG. 35 is a flowchart showing the operation of the command queue control unit 3322 in FIG. FIG. 36 is a flowchart showing the operation of the command sending unit 3323 in FIG. The non-priority data output operation will be described below with reference to FIGS.

  As shown in FIG. 34, when receiving a command requesting reading of non-priority data from the user terminal 3208 (step S101), the command receiving unit 3327 is free in the command queue 3321 corresponding to the user terminal 3208 that transmitted the command. It is determined whether or not there is (step S102). If there is no available command queue 3321, the command receiving unit 3327 rejects the command reception (step S103) and returns to the operation of step S101. On the other hand, if the corresponding command queue 3321 has a vacancy, the command reception unit 3327 stores the received command in the last part of the corresponding command queue 3321 (step S104), and ends the operation. The non-priority input / output control unit 3302 does not need to secure resources for sending data to the user terminal 3208 unlike the above-described priority input / output control unit 3301.

  As shown in FIG. 35, the command queue control unit 3322 refers to the management information stored in the buffer management unit 3325, and from the command queue 3321 for each user terminal 3208 in which commands are stored, data The amount of data stored in the buffer unit 3324 (including not only the amount of data actually stored but also the amount of data reserved for storage) is the smallest (including the case where no data is stored at all). ) A command queue corresponding to the user terminal 3208 is selected (step S201). When there are a plurality of user terminals having the same data amount stored in the data buffer unit 3324 and each having the smallest data amount, the queue length is the longest (that is, the largest number of commands). Command queue) is selected. Further, in this case, when there are a plurality of command queues having the same queue length, the command queues are cyclically selected in the order determined by default.

  Next, the command queue control unit 3322 pays attention to the selected command queue, and sets the command stored at the head of the command queue to COM (1) (step S202). Next, the command queue controller 3322 sets an initial value 1 to the command counter Nc and sets an initial value 0 to the sending command counter Ns (step S203). Next, the command queue control unit 3322 determines whether the destination MSFS 3206 of the command COM (Nc) is equal to the destination MSFS 3206 of the command COM (1) (step S204). If the MSFSs 3206 that are the destinations of both commands are equal, the command queue control unit 3322 extracts the command COM (Nc) from the selected command queue, and outputs the extracted command COM (Nc) to the command transmission unit 3323 (step S205). . Since the initial value of the count value of the command counter Nc is 1, the command COM (Nc) = COM (1) is initially set. Therefore, at the beginning, the head command COM (1) is extracted from the selected command queue, and is output to the command transmission unit 3323.

  Next, the command queue control unit 3322 increments the send command counter Ns (step S206). Next, the command queue control unit 3322 determines whether or not the count value of the send command counter Ns is smaller than a predetermined threshold value N (step S207). Here, the count value of the send command counter Ns corresponds to the number of commands sent by the command queue control unit 3322 to one MSFS 3206 from the selected command queue. When a large number of commands are sent to one MSFS 3206 almost simultaneously, congestion occurs in the MSFS 3206 that has received the command. In this embodiment, N or more commands are not sent from one command queue 3321 at the same time. I am doing so. That is, the command queue control unit 3322 can proceed to the transmission process for the next command in the selected command queue only when the count value of the transmission command counter Ns is smaller than the predetermined threshold value N.

  When the count value of the send command counter Ns is smaller than the predetermined threshold value N in step S207, the command queue control unit 3322 has a command indicated by COM (Nc + 1) in the selected command queue. It is determined whether there is a command to be transmitted in the selected command queue (step S208). When the command COM (Nc + 1) exists in the selected command queue, that is, when the command to be sent still exists in the selected command queue, the command queue control unit 3322 increments the command counter Nc (step S209). ), Returning to the operation of step S204. Therefore, the command stored at the next position in the selected command queue is the target of the transmission process to the command transmission unit 3323. For example, when considering the command COM (2) stored at the second position in the selected command queue, if this command COM (2) is equal to the command COM (1) at the head position, the command COM (2) is output to the command sending unit 3323 in step S205. On the other hand, when the command COM (2) is not equal to the command COM (1) at the head position, steps S205 to S207 are skipped after step S204, and the command COM (2) is not output to the command sending unit 3323. .

  On the other hand, when the command COM (Nc + 1) does not exist in the selected command queue, the command queue control unit 3322 determines whether the command transmission processing is completed, that is, the data corresponding to the command output to the command transmission unit 3323 is MSFS 3206. Is determined whether it has arrived at the data buffer unit 3324 (step S210). If completed, the process returns to step S201. Note that the command queue control unit 3322 determines that the command transmission processing has been completed when receiving a reception completion notification (see step S308 in FIG. 36 described later) from the command transmission unit 3323.

  If the count value of the send command counter Ns is equal to or greater than the predetermined threshold value N in step S207 described above, the command queue control unit 3322 proceeds to the operation in step S210, and sends processing for the next command. Do not execute.

  As shown in FIG. 36, when receiving a command requesting to read non-priority data from the command queue control unit 3322, the command sending unit 3323 initializes a read completion counter Nr (step S301). That is, the command sending unit 3323 sets the total number of commands received from the command queue control unit 3322 as an initial value in the read completion counter Nr. Next, the command sending unit 3323 determines whether or not it holds a command not sent to the MSFS 3206 (step S302). If the command sending unit 3323 has an unsent command, the command sending unit 3323 selects one command (usually the next command) from the held unsent commands and pays attention to the command. When the data corresponding to is read from the storage device 3205, it is determined whether or not there is a free area in the data buffer unit 3324 that can store the data (step S303). Whether there is an empty area in the data buffer unit 3324 can be determined by referring to the management information (specifically, the count value of each counter) in the buffer management unit 3325.

  When there is an empty area in the data buffer unit 3324, the command sending unit 3323 reserves an area for storing data in the data buffer unit 3324 and notifies the buffer management unit 3325 that the area has been reserved. The selected command is sent to the destination MSFS 3206 (step S304). If reservation is not possible, the command sending unit 3323 does not send a command requesting reading of non-priority data to the MSFS 3206 until a reservation is possible. By performing this processing, it is possible to prevent data from being discarded because there is no storage area in the data buffer unit 3324 when data is received from the MSFS 3206.

  The buffer management unit 3325 has a counter for each user terminal, and when receiving a reservation notification from the command transmission unit 3323, the user terminal 3208 that has transmitted a data read command for reserving a storage area to the data buffer unit 3324. The counter value corresponding to is incremented.

  Next, the command sending unit 3323 determines whether the data buffer unit 3324 has received data from the MSFS 3206 (step S305). When the data buffer unit 3324 has not received the data from the MSFS 3206, the command sending unit 3323 returns to the operation of step S302 and performs sending processing for the next unsent command. On the other hand, when the data buffer unit 3324 receives data from the MSFS 3206, the command sending unit 3323 decrements the read completion counter Nr (step S306). Therefore, the count value of the read completion counter Nr is the command for which data transmission from the MSFS 3206 has not been completed among the commands to be sent at this time (that is, all commands received from the command queue control unit 3322). It represents a number.

  Thereafter, the command sending unit 3323 determines whether or not the count value of the read completion counter Nr has become 0 (step S307). If the count value of the read completion counter Nr is not 0, the command sending unit 3323 returns to the operation of step S302 and performs sending processing for the next unsent command. On the other hand, when the count value of the read completion counter Nr becomes 0, the command sending unit 3323 sends a reception completion notification to the command queue control unit 3322 (step S308), and the operation ends.

  As is clear from the above description, after sending all the commands sent from the command queue control unit 3322, the command sending unit 3323 continues to receive the data corresponding to the sent commands from the MSFS 3206 until the next reception is completed. The command is not received from the command queue control unit 3322.

  The MSFS 3206 that has received a command requesting reading of non-priority data temporarily stores the command in a non-priority command queue provided therein. Then, the MSFS 3206 retrieves a command from the non-priority command queue and reads data from the storage device 3205 within the read band secured as the non-priority resource. The read data is transmitted using a communication band as a non-priority resource that is secured with the MSFS 3206 by the I / O unit 3207 that has requested reading. At this time, since the I / O unit 3207 does not communicate with a plurality of MSFSs 3206 at the same time with respect to reception of non-priority data, the MSFS 3206 has the communication bandwidth of the non-priority resources that the I / O unit 3207 has. Data can be transmitted using all of them.

  When data from the MSFS 3206 arrives at the I / O unit 3207, the data is stored in an already reserved area in the data buffer unit 3324. The data sending unit 3326 takes out data from the data buffer unit 3324 within a communication band with the outside of the server device secured as a non-priority resource, and sets an area where the data is stored as a free area. At the same time, the data transmission unit 3326 notifies the buffer management unit 3325 of data extraction. Receiving the notification, the buffer management unit 3325 decrements the counter corresponding to the user terminal 3208 that has transmitted the read command for the data extracted from the data buffer unit 3324. Then, the data sending unit 3326 sends the extracted data to the corresponding user terminal 3208.

  As described above, according to the eighth embodiment, the resources of each part of the system are the resources used to output priority data (priority resources) and the resources used to output non-priority data ( Therefore, the resource used for reading the non-priority data can be secured independently of the priority resource. As a result, even if the read amount of the non-priority data increases, the priority data output is not affected.

  By the way, there is a limit to the ability of each user terminal 3208 to receive non-priority data from the server device. Nevertheless, when a certain user terminal 3208 requests reading of non-priority data in an amount exceeding its own receiving capability, the data read from the storage device 3205 in response to the request from the user terminal 3208 is A large amount of data is accumulated in the data buffer unit 3324. As a result, in the server device, processing for read requests from other user terminals 3208 is delayed. Hereinafter, this will be described in more detail with a specific example.

  For example, it is assumed that ten user terminals 3208 are connected to the server device, each user terminal 3208 has a receiving capability of 10 Mbps, and the communication bandwidth with the outside of the server device is 100 Mbps as a whole. To do. In this case, if a certain user terminal 3208 issues a read request for non-priority data exceeding 100 Mbps to the server device, the data buffer unit 3324 has one user terminal that has issued the read request. The stored data for 3208 is full. In addition, the data stored in the data buffer unit 3324 at this time can only be reduced at a rate of 10 Mbps. Under such circumstances, even if there is a read request for non-priority data from another user terminal 3208, the other user terminal 3208 is not available until there is a free space in the data buffer unit 3324 for storing read data. The transmission process for the request from the server cannot be executed. This makes it impossible to quickly respond to a read request command that requires immediacy. Moreover, in this case, although the server apparatus as a whole has a communication capability of 100 Mbps, it actually exhibits only a communication capability of 10 Mbps, and effectively uses a limited communication band. I can't.

  Therefore, in the eighth embodiment, by referring to the management information stored in the buffer management unit 3325, the command buffer 3321 for each user terminal 3208 in which commands are stored is stored in the data buffer unit 3324. Since the command queue corresponding to the user terminal 3208 with the smallest amount of data stored is selected and the process of sending the commands stored in the selected command queue is preferentially executed, a plurality of users Non-priority resources that are commonly used by the terminal 3208 are less occupied and used by one user terminal 3208, and the server device can make maximum use of the communication capability of each user terminal 3208. Can do.

  Note that the command queue control unit 3322 may stop taking out a command from the selected command queue 3321 when the smallest amount of data stored in the data buffer unit 3324 exceeds a predetermined threshold value. Good. As a result, it is possible to more effectively prevent the data buffer unit 3324 from being filled with the data of the specific user terminal 3208 and the non-prioritized data read request of the other user terminals 3208 from being processed.

  In the eighth embodiment, the command sending unit 3323 reserves an area for storing data in the data buffer unit 3324, and then sends a command requesting reading of non-priority data to the MSFS 3206. ing. That is, if the storage area cannot be reserved, the command transmission unit 3323 does not transmit a command requesting reading of non-priority data to the MSFS 3206 until the storage area can be reserved. By providing such a command sending unit 3323, it is possible to prevent data from being discarded because there is no storage area in the data buffer unit 3324 when data is received from the MSFS 3206.

  Furthermore, in the eighth embodiment, after sending all the commands sent from the command queue control unit 3322, the command sending unit 3323 continues to receive the data corresponding to the sent commands from the MSFS 3206 until the next reception is completed. The command is not received from the command queue control unit 3322. Thus, the I / O unit 3207 does not communicate with a plurality of MSFSs 3206 at the same time with respect to reception of non-priority data, so there is no need to consider cell collision between non-priority data. Further, the MSFS 3206 can transmit data at high speed using all the communication bands of the non-priority resources that the I / O unit 3207 has.

(Ninth embodiment)
FIG. 37 is a block diagram showing a configuration of a data transmission system using a server apparatus according to the ninth embodiment of the present invention. 37, the data transmission system includes a resource management unit 3701, a data storage unit 3702, an exchange unit 3703, an input / output unit 3704, and a plurality of user terminals 3708. Note that the server device of this embodiment includes a resource management unit 3701, a data storage unit 3702, an exchange unit 3703, and an input / output unit 3704. The server device and each terminal 3708 are connected via a wired or wireless transmission path.

  The resource management unit 3701 manages necessary resources when sending data to each user terminal 3708. The data storage unit 3702 includes a plurality of storage devices (for example, an optical disc driver with an autochanger) 3705 capable of random access, and a plurality of MSFSs 3706 connected to each storage device 3705 and controlling reading of data. Exchanger 3703 exchanges data between input / output unit 3704 and data storage unit 3702. In this embodiment, as an example, an ATM switch is used for the exchange unit 3703. The input / output unit 3704 includes a plurality of I / O units 3707 and controls input / output of data between the server apparatus and each user terminal 3708. Each user terminal 3708 is connected to one of the I / O units 3707.

  In the configuration as described above, a connection is set in a mesh shape by the exchange unit 3703 between each I / O unit 3707 and each MSFS 3706. Note that the resource management unit 3701 manages the bandwidth of each connection.

The ninth embodiment is different from the eighth embodiment described above in the following points.
First, the resource management unit 3701 manages the resources of the entire system separately into priority resources and non-priority resources. However, the amount allocated as the priority resource and the non-priority according to the increase / decrease in the required priority resource amount Change the amount allocated as a resource.

  Next, a communication band as a non-priority resource of the ATM switch in the exchange unit 3703 is secured by the ABR mode. Here, ABR is a service category deliberated by "The ATM Forum", which is an industry group that promotes standardization of communication in ATM. In communication using ABR, when congestion occurs, the transmission rate is automatically adjusted at the ATM layer, thereby eliminating the congestion. Further, when there is a vacancy in the band, the transmission rate is automatically increased, and the communication band is effectively used. ABR is described in ATM Forum Traffic Management Specification Version 4.0, October 1995, ATM Forum / 95-0013R8, and the like.

  FIG. 38 is a block diagram showing a more detailed configuration of the I / O unit 3707 in FIG. In FIG. 38, the I / O unit 3707 includes a priority input / output control unit 3801 and a non-priority input / output control unit 3802. Non-priority input / output control unit 3802 includes a command queue 3841, a command queue control unit 3842, a command sending unit 3843, a data buffer unit 3844, a data sending unit 3848, and a command receiving unit 3847. The command sending unit 3843 includes a plurality of counters 3848 and a plurality of sending queues 3849 provided for each MSFS 3706.

  A priority input / output control unit 3801 controls input / output of priority data such as moving image data. On the other hand, the non-priority input / output control unit 3802 controls input / output of non-priority data such as still image data.

  In the non-priority input / output control unit 3802, the command queue 3841 collectively stores commands that request reading of non-priority data from the user terminal 3708. The command queue control unit 3842 controls the extraction of commands from the command queue 3841. The command sending unit 3843 sends the command extracted by the command queue control unit 3842 to the corresponding destination MSFS 3706. The data buffer unit 3844 stores data read by the MSFS 3706 from the storage device 3705. The data sending unit 3846 sends the data stored in the data buffer unit 3844 to the corresponding destination user terminal 3708. The command reception unit 3847 manages whether or not a command that has arrived from the user terminal 3708 is registered in the command queue 3841.

  In the command transmission unit 3843, each counter 3848 counts the number of commands for which data has not yet arrived from the MSFS 3706 among the transmitted data read commands for each MSFS 3706. Each transmission queue 3849 stores a command to be transmitted for each MSFS 3706.

  FIG. 39 is a flowchart showing the operation of the resource management unit 3701 in FIG. FIG. 40 is a flowchart showing the operation of the command receiving unit 3847 in FIG. FIG. 41 is a flowchart showing the operation of the command queue controller 3842 in FIG. FIG. 42 is a flowchart showing the operation of the command sending unit 3843 in FIG. Hereinafter, the operation of the server device of the ninth embodiment will be described with reference to FIGS. 39 to 42.

  First, the resource initial setting operation performed when the system is started will be described. At the time of system startup, the resource management unit 3701 is responsible for the resources of the entire data transmission system (data read bandwidth of each MSFS 3706, communication bandwidth between the I / O unit 3707 and the outside, communication of each ATM switch in the exchange unit 3703 Bandwidth) to be secured as a minimum non-priority resource. In this embodiment, the priority resource is always secured in preference to the non-priority resource. Therefore, unless the minimum amount of bandwidth to be secured as a non-priority resource is determined, when priority data read requests increase, there is no bandwidth secured as a non-priority resource, and non-priority data cannot be transmitted. . Further, at the time of system startup, the resource management unit 3701 sets an ATM communication path as a non-priority resource between the MSFS 3706 and the I / O unit 3707 in the ABR mode.

  Next, priority data output operation will be described. When each user terminal 3708 requests the server device to output priority data, the user terminal 3708 requests to secure resources of a band necessary for transmitting the priority data. The priority input / output control unit 3801 that has received the request for securing the resource requests the resource management unit 3701 to secure the resource by specifying the bandwidth amount. In order to transmit the requested bandwidth amount of data, the resource management unit 3701 uses the priority switch resource 3703, the data read bandwidth of the MSFS 3206, and the communication bandwidth of the I / O unit 3707 as priority resources. Is secured from the bandwidth amount allocated in advance.

  If all or most of the bandwidth that is currently reserved as a priority resource is in use and there is not enough bandwidth to read and send the requested priority data, as shown in FIG. The resource management unit 3701 determines that a request for changing the amount of bandwidth secured as a priority resource has occurred (step S401). Thereafter, if the resource management unit 3701 is changed so as to increase the bandwidth amount of the priority resource, the above-mentioned minimum bandwidth amount (set at the time of starting the system) can be secured for the non-priority resource. Whether or not (step S402). If the bandwidth amount of the priority resource is increased and the minimum bandwidth amount cannot be secured for the non-priority resource, the resource management unit 3701 rejects the increase of the priority resource (step S403), and the operation of step S401 Return to. In this case, securing the priority resource fails, and the server apparatus cannot execute the output of the priority data.

  On the other hand, when the minimum bandwidth can be secured for the non-priority resource even if the bandwidth amount of the priority resource is increased, the resource management unit 3701 displays the non-priority resource allocated as the remaining priority resource after the change. The bandwidth amount is set in the MSFS 3706 and the I / O unit 3707, and the threshold value L of the counter 3848 is notified to the command transmission unit 3843 (step S404). Note that, when the priority input / output control unit 3801 completes the output of the priority data, the resource management unit 3701 decreases the priority resource that secures the corresponding bandwidth amount.

  Other operations of the resource management unit 3701 are the same as those of the resource management unit 3201 shown in FIG. 32, and a description thereof will be omitted.

  Next, the non-priority data output operation will be described. As illustrated in FIG. 40, when receiving a command requesting reading of non-priority data from the user terminal 3708 (step S501), the command receiving unit 3847 determines whether or not the command queue 3841 has an empty space (step S502). ). If there is no free space in the command queue 3841, the command reception unit 3847 rejects the reception of the command to the user terminal 3708 that transmitted the command (step S503), and returns to the operation of step S501. On the other hand, when the command queue 3841 has a vacancy, the command reception unit 3847 stores the received command in the last part of the command queue 3841 (step S504), and ends its operation.

  As shown in FIG. 41, the command queue control unit 3842 determines whether or not a command is stored in the command queue 3841 (step S601). If it is stored, the transmission queue 3849 provided for each MSFS. In step S602, it is determined whether or not there is a queue in which at least one of the queues is full (that is, a state where commands are stored up to the end). If there is at least one queue that is full, the command queue control unit 3842 stops taking out commands from the command queue 3841 until a command is transmitted and the transmission queue 3849 becomes empty. On the other hand, if there is no full send queue 3849, the command queue control unit 3842 takes out one command from the head of the command queue 3841 and stores the command in the send queue 3849 corresponding to the destination MSFS 3706 ( Step S603).

  Next, the command queue control unit 3842 determines whether or not a command corresponding to a predetermined data amount has been sent to the command sending unit 3843 (step S604). Here, the predetermined data amount is a non-priority resource allocated to the I / O unit 3707 at that time, and includes a communication band between the MSFS 3706 and a communication band with the user terminal 3708. It is defined as the amount of data that can be transmitted during a predetermined time T using the smaller communication band. When the command corresponding to the predetermined data amount has not been sent to the command sending unit 3843, the command queue control unit 3842 returns to the operation of step S601, takes out the next command from the command queue 3841, and sends it to the command sending unit 3843. To do. On the other hand, when a command corresponding to a predetermined data amount is sent to the command sending unit 3843, the command queue control unit 3842 stops sending commands from the command queue 3841 to the command sending unit 3843 for a time T. (Step S605).

  For example, the communication band as a non-priority resource for the I / O unit 3707 to output data to the outside of the server apparatus is M (KB / s), and the I / O unit 3707 is for receiving read data from the MSFS 3706 When the communication bandwidth as the non-priority resource is N (KB / s), the amount of data read by one read request command is S (KB), and M> N, the command queue control unit 3842 A command is extracted from the command queue 3841 at a cycle of / N (sec) and sent to the command sending unit 3843.

  As shown in FIG. 42, the command sending unit 3843 initializes a value i representing the command sending destination MSFS, that is, sets it to 1 (step S701). Next, the command transmission unit 3843 determines that the count value Co (i) of the i-th counter 3848 is greater than the threshold value L (the threshold value set in step S404 described above) set by the resource management unit 3701. It is determined whether or not it is smaller (step S702). When the count value Co (i) is smaller than the threshold value L (that is, when Co (i) <L), the command transmission unit 3843 increments the count value Co (i) of the i-th counter 3848 ( In step S703), one command is taken out from the i-th sending queue 3849 and sent to the corresponding MSFS 3706 (step S704). Next, the command sending unit 3843 increments the value i (step S705).

  Thereafter, the command sending unit 3843 determines whether or not the data buffer unit 3844 has received read data from the MSFS 3706 (step S706). The count value of 3848 is decremented (step S707). Next, the command transmission unit 3843 determines whether or not the value i is larger than the total number of MSFS 3706 (step S708). If the value i is less than or equal to the total number of MSFSs 3706, the command sending unit 3843 returns to the operation of step S702 and executes command sending processing for the next sending queue 3849.

  Here, in step S702 described above, when the count value Co (i) of the i-th counter 3848 is greater than or equal to the threshold value L (that is, when Co (i) ≧ L), the command transmission unit 3843 responds. Command transmission processing for the i-th transmission queue 3849 is not performed. This is to avoid the concentration of read processing for one MSFS 3706 and the retention of commands in the MSFS 3706.

  If the value i is larger than the total number of MSFS 3706 in step S708, the command sending unit 3843 returns to the operation in step S701. Therefore, the value i is set to the initial value 1, and the above series of operations is repeated. As described above, the command sending unit 3843 searches the plurality of sending queues 3849 in order from the first, and when the sending queue 3849 of interest stores a command, only one command is taken out and handled. The data is sent to the destination MSFS 3706. As a result, each MSFS 3706 performs the reading process evenly, and it is possible to avoid the concentration of the load on any one of the MSFSs 3706. Furthermore, even if a read request from the user terminal 3708 is accidentally biased to one MSFS 3706, if an unprocessed command (command for which no data has been read) for the MSFS 3706 becomes equal to or greater than the threshold value L. Since the command sending process to the MSFS 3706 is stopped, it is possible to cope with such an unexpected situation.

  The MSFS 3706 that has received the non-priority data read request temporarily stores the request in the internal non-priority command queue. Then, the MSFS 3706 retrieves a command from the non-priority command queue and executes data reading within the read band secured as the non-priority resource.

  The read data is transmitted as a non-priority resource using the communication band secured in the ABR mode by the I / O unit 3707 that requested the reading with the MSFS 3706.

  When data from the MSFS 3706 arrives at the I / O unit 3707, it is stored in the data buffer unit 3844. At this time, the count value of the counter 3848 corresponding to the transmission source MSFS 3706 is decremented (step S707 described above).

  The data transmission unit 3846 extracts data from the data buffer unit 3844 according to the communication band with the outside of the server device secured as a non-priority resource, and transmits the extracted data to the corresponding user terminal 3708.

  As described above, according to the ninth embodiment, when resources that are not used as priority resources are all managed as non-priority resources and it is necessary to newly increase the priority resources, Since priority resources are allocated from non-priority resources under conditions that do not fall below the prescribed resource amount (minimum bandwidth), priority data input / output is affected even if non-priority data read increases. In addition, since all unused resources are used as non-priority resources, reading of non-priority data is performed efficiently.

  In the ninth embodiment, since each MSFS 3706 operates asynchronously, collision of data transmitted from each MSFS 3706 to the I / O unit 3707 may occur, and congestion may occur in the ATM switch in the switching unit 3703. is there. However, in the ninth embodiment, the ATM switch communication band as a non-priority resource is secured using the ABR mode, and the MSFS 3706 that has read the data transmits data using the communication band secured in the ABR mode. Therefore, even if congestion occurs, the transmission rate from the MSFS 3706 immediately decreases, and the congestion is eliminated. Therefore, the I / O unit 3707 can always receive data normally. However, when a free band is generated in the communication band due to the release of the communication band as the priority resource, the transmission rate immediately increases and the band is effectively used.

  In the ninth embodiment, the command queue control unit 3842 includes a communication band as a non-priority resource for the I / O unit 3707 to output data to the outside of the server device, and the I / O unit 3707 from the MSFS 3706. The commands are sequentially extracted from the head of the command queue 3841 in a cycle in which the data amount requested by the data read command is equal to the smaller one of the communication bands as non-priority resources for receiving the read data. The command is sent to the command sending unit 3843. As a result, the average bandwidth used by the data sent from the MSFS 3706 becomes equal to or smaller than the communication bandwidth as a non-priority resource for the ATM switch to communicate with the MSFS 3706, so that the data of the MSFS 3706 An effect of suppressing congestion of the ATM switch due to transmission can be expected. Furthermore, since the average bandwidth used by the data sent from the MSFS 3706 is equal to or smaller than the communication bandwidth as a non-priority resource for outputting data to the outside of the server device, the I / O An increase in the amount of data stored in the data buffer unit 3844 in the unit 3707 can be suppressed.

  In the ninth embodiment, the command sending unit 3843 receives the attention only when the count value of the counter 3848 corresponding to the sending queue 3849 of interest is less than a predetermined threshold value L. Since the command in the sending queue 3849 is transmitted to the MSFS 3706, the number of data that can be transmitted simultaneously from the MSFS 3706 is limited, and the effect of suppressing congestion of the ATM switch is achieved. I can expect.

  In the ninth embodiment, the MSFS 3706 extracts a command from the internal non-priority command queue and executes data reading within the read bandwidth reserved as the non-priority resource. When I / O processing is not being performed, a command may be extracted from the non-priority command queue and non-priority data may be read. In this case, it is not necessary to notify each MSFS 3706 of the read bandwidth of data as a non-priority resource when the bandwidth amount of the non-priority resource changes.

  In the ninth embodiment, the counter 3848 and the sending queue 3849 are provided for each MSFS 3706. However, the counter 3848 and the sending queue 3849 are provided for each I / O unit 3707 so that the destination MSFS is not conscious. However, as in the ninth embodiment, the number of data that may be transmitted simultaneously from the MSFS 3706 is limited, and an effect of suppressing ATM switch congestion can be expected. However, when the reading of data from the user terminal 3708 is concentrated to a certain MSFS 3706 to some extent, the reading efficiency may be deteriorated. For MSFS 3706 where the request is not concentrated, a response should be obtained as soon as the request can be transmitted. However, if data is not received from the MSFS where the request is concentrated, the counter 3848 counts. This is because the request cannot be transmitted to the MSFS 3706 by the threshold value control.

  Generally, a user of a file system designates a range of file data to be acquired and sends a data transmission request to a server in the file system. The server sends all file data in the range specified by the data send request to the user. In the conventional file system, the user can perform an operation to interrupt the data transmission process being executed by the server, but the user temporarily stops the data transmission process being executed and resumes the suspended data transmission process. I couldn't.

  By the way, in a file system that handles digitized moving image data, continuous data transmission processing at an arbitrary data transmission rate is indispensable. Regarding the file system that handles moving image data, the draft of ISO / IEC standard 13818-9 defines a pause and restart operation for continuous transmission of data related to one moving image file. According to this draft, it is defined that data transmission is temporarily stopped by “DSM Stream Pause” and data transmission is resumed by “DSM Stream Resume”.

  However, in a conventional file system that handles moving image data, data that is arbitrarily specified at the time of pause release regarding pause that pauses continuous transmission processing of data at a certain transmission speed and pause release that resumes paused data transmission processing. A pause control method for resuming data transmission at a transmission rate has not been established.

  Further, the conventional file system does not consider the time required for data acquisition after the pause is released when the data transmission is resumed by the pause release. Therefore, if the data transmission speed changes before and after the pause process, the data required for continuous transmission will be insufficient, the acquired data will overflow from the buffer, or the file system will continue to the next data before data transmission resumes. There were problems such as requiring time to acquire.

  Therefore, in a file system that suspends and resumes continuous transmission processing of data, even if the transmission rate after pause release is arbitrarily specified, there is no data shortage at the time of data transmission, and the time until restart is reduced. Embodiments that can be shortened are described below.

(Tenth embodiment)
FIG. 43 is a block diagram showing a configuration of a file system according to the tenth embodiment of the present invention. In FIG. 43, this file system includes a server device 4301 and a data storage device 4302.

  The data storage device 4302 stores data managed by the server device 4301 and used by the user 4303. The user 4303 sends a data transmission request for requesting transmission of data in a specified range, a pause request for requesting temporary suspension of data transmission, and a pause release request for requesting resumption of temporary data transmission. Issued to 4301. The server device 4301 receives a data transmission request from the user 4303, acquires data from the data storage device 4302, and transmits the data to the user 4303. The data transmission unit 4310 and the data acquisition unit 4311 A buffer 4312, a buffer management unit 4313, and a pause control unit 4314.

  The data acquisition unit 4311 acquires data from the data storage device 4302 sequentially from the specified position and continuously at the specified speed. The buffer 4312 stores data acquired from the data storage device 4302. The buffer management unit 4313 operates the data stored in the buffer 4312 and manages the data amount. The data sending unit 4310 sends the data stored in the buffer 4312 to the user 4303 continuously in the designated order and at the designated arbitrary transmission rate. In response to a pause request and a pause release request from the user 4303, the pause control unit 4314 controls the data transmission unit 4310, the data acquisition unit 4311, and the buffer management unit 4313. In addition, all data handled by the server device 4301 and the data storage device 4302 are operated in units of fixed-length blocks, and block numbers indicating positions in the data are continuously given to the blocks.

  FIG. 44 is a time chart showing the relationship among the data transmission request, the data acquisition request, and the data transmission in the data transmission operation of the server apparatus 4301. First, with reference to FIG. 44, the operation when the server apparatus 4301 receives a data transmission request from the user 4303 will be described.

  In FIG. 44, reference numerals 4441 and 4442 indicate a time axis, and the corresponding positions in the vertical direction indicate the same time. Reference numeral 4431 indicates a data transmission request received from the user 4303 by the server apparatus 4301 and its reception time. Reference numbers 4401A to 4403A indicate a data acquisition request issued by the data acquisition unit 4311 to the data storage device 4302 and its issue time. Reference numbers 4411A to 4413A indicate the data acquired by the data acquisition requests 4401A to 4403A and the time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4421 indicates the maximum time from when the data acquisition request 4401A is issued until the corresponding data 4411A is stored in the buffer 4312. Reference numbers 4451 to 4454 indicate time intervals at which the data sending unit 4310 sends data to the user 4303.

  The user 4303 issues a data transmission request 4431 and designates a data range and a transmission speed at the time of transmitting the data. The data range is specified by a start block number and an end block number. The data transmission rate is specified by a data transmission interval (hereinafter referred to as Tsd) which is an average time interval between blocks to be transmitted.

  Upon receiving the data transmission request 4431, the data acquisition unit 4311 issues a data acquisition request 4401A for acquiring one block corresponding to the start block number specified by the data transmission request 4431. The data acquisition unit 4311 issues a data acquisition request 4402A for acquiring the next one block to the data storage device 4302 after the data transmission interval Tsd from the issue time of the data acquisition request 4401A. By the same procedure, the data acquisition unit 4311 repeatedly issues a data transmission request until the end block number designated by the data transmission request 4431 is reached.

  The block acquired from the data storage device 4302 by the data acquisition request 4401A is stored in the buffer 4312. The data acquisition time D is the maximum time from when the data acquisition unit 4311 issues a data acquisition request to when the corresponding block is stored in the buffer 4312, and is a known value. That is, it is guaranteed that one specified block can be acquired during the data acquisition time D.

  The block acquired by the data acquisition request 4401A and stored in the buffer 4312 is transmitted to the user 4303 by the data transmission unit 4310 at the data transmission time 4411A after the data acquisition time D has elapsed from the issue time of the data acquisition request 4401A. Is done. Similarly, the block acquired by the data acquisition request 4402A is transmitted to the user 4303 by the data transmission unit 4310 at the transmission time 4412A after the data acquisition time D has elapsed from the issue time of the data acquisition request 4402A. By the same procedure, data transmission is continuously performed until data corresponding to the end block number specified by the data transmission request 4431 is transmitted. With the above operation, continuous data transmission at an arbitrary data transmission rate designated by the user 4303 is realized.

  FIG. 45 is a time chart showing the relationship among the data transmission request, the data acquisition request, and the data transmission in the pause operation of the server apparatus 4301. Next, an operation when the pause control unit 4314 in the server apparatus 4301 receives a pause request from the user 4303 will be described with reference to FIG.

  In FIG. 45, reference numeral 4432 indicates a pause request received from the user 4303 by the pause control unit 4314 and its reception time. Reference numbers 4404A to 4406A indicate a data acquisition request issued by the data acquisition unit 4311 to the data storage device 4302 and its issue time. Reference numbers 4414A and 4415A indicate the data acquired by the data acquisition requests 4044A and 4045A, and the time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4422 indicates the maximum time D from when the data acquisition request 4404A is issued until the corresponding data 4414A is stored in the buffer 4312. Reference numerals 4455 to 4457 indicate time intervals at which the data sending unit 4310 sends data to the user 4303.

  In response to the data transmission request from the user 4303, when the data acquisition unit 4311 is about to issue the data acquisition requests 4404A to 4406A and after the data transmission requests 4404A and 4405A are issued, the pause control unit 4314 When the pause request 4432 is received, the pause control unit 4314 controls the data acquisition unit 4311 to immediately stop issuing the data acquisition request. As a result, the data acquisition request 4406A is not issued from the data acquisition unit 4311. Similarly, the pause control unit 4314 controls the data sending unit 4310 to immediately stop sending data. Since the transmission time of the blocks 4414A and 4415A acquired by the data acquisition requests 4404A and 4405A is later than the pause request reception time, the blocks 4414A and 4415A remain stored in the buffer 4312, and the data transmission unit 4310 Will not be sent to the user 4303. With the above operation, the temporary stop of the data that has been continuously transmitted is realized.

  FIG. 46 is a time chart showing the relationship among the data transmission request, the data acquisition request, and the data transmission in the pause release operation of the server apparatus 4301. Next, with reference to FIG. 46, an operation when the pause control unit 4314 receives a pause release request from the user 4303 will be described.

  In FIG. 46, reference numeral 4433 indicates a pause release request received from the user 4303 by the pause control unit 4314 and its reception time. Reference numbers 4406A to 4408A indicate a data acquisition request issued by the data acquisition unit 4311 to the data storage device 4302 and its issue time. Reference numbers 4414A and 4415A indicate the two blocks that are still stored in the buffer in the pause process, and the transmission time to the user 4303. Reference numbers 4416A and 4417A indicate data acquired by the data acquisition requests 4406A and 4407A after the pause release, and the time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4423 indicates the maximum time D from when the data acquisition request 4406A is issued until the corresponding data 4416A is stored in the buffer 4312. Reference numerals 4458 to 4462 indicate time intervals at which the data sending unit 4310 sends data to the user 4303.

  The user 4303 issues a pause release request 4433 and designates the data transmission rate after the pause release. The data transmission rate is specified by a data transmission interval Tsd ′ that is an average time interval between blocks to be transmitted, as in the case of data transmission request.

  When the pause control unit 4314 receives the pause release request 4433 from the user 4303, the pause control unit 4314 refers to the buffer management unit 4313 to obtain the number N of blocks that are stored in the buffer 4312 after the pause process. .

  In response to the pause release request 4433, a time at which the pause control unit 4314 controls the data acquisition unit 4311 to immediately resume data acquisition is defined as a data acquisition restart time trd. Similarly, the transmission time of the next block to be transmitted among the blocks stored in the buffer 4312, that is, the data transmission time 4414A in FIG. 46 is defined as the data transmission resumption time tsd.

The pause control unit 4314 substitutes the data acquisition restart time trd, the number N of blocks in the buffer 4312, the data transmission interval Tsd ′ after the pause release, and the data acquisition time D into the following equation (14). The data transmission restart time tsd is obtained.
tsd = trd + DN · Tsd ′ (14)

  Next, the pause control unit 4314 compares the data transmission resumption time tsd obtained using the above equation (14) with the data acquisition resumption time trd, and the data transmission resumption time tsd is equal to or greater than the data acquisition resumption time trd. In this case, the time determined by the equation (14) is determined as the data transmission restart time tsd. If the data transmission resumption time tsd is less than the data acquisition resumption time trd, a time equal to the data acquisition resumption time trd is determined as the data transmission resumption time tsd.

  Next, the pause control unit 4314 starts from the data acquisition request 4406A for acquiring the block 4416A next to the last block 4415A stored in the buffer 4312 to the data acquisition unit 4311 at the data acquisition restart time trd. Then, control is performed to issue a data acquisition request until the end block is reached. Similarly, the pause control unit 4314 transmits data to the data transmission unit 4310 at the determined data transmission restart time tsd until the end block is transmitted in order from the first block 4414A stored in the buffer 4312 immediately. Control to send.

  As described above, according to the tenth embodiment, when an arbitrary transmission rate is designated as the data transmission rate after the pause release, the pause control unit 4314 resumes data transmission using the above-described equation (14). By obtaining the time, the data can be transmitted at a designated transmission speed without interruption after the pause is released.

(Eleventh embodiment)
The file system according to the eleventh embodiment of the present invention will be described below with reference to the drawings. The eleventh embodiment differs from the tenth embodiment shown in FIG. 43 in the function of each block, but the overall configuration is the same. Therefore, the eleventh embodiment will be described below with the aid of the configuration of the tenth embodiment shown in FIG.

  In the eleventh embodiment, the operation when the server apparatus 4301 receives a data transmission request from the user 4303 is the same as the operation described with reference to FIG. 44 in the tenth embodiment. Similarly, the operation when the pause control unit 4314 receives a pause request from the user 4303 is the same as the operation described with reference to FIG. 45 in the tenth embodiment. Therefore, description of these operations is omitted.

  FIG. 47 is a time chart showing the relationship among the data transmission request, the data acquisition request, and the data transmission in the pause release operation of the server apparatus 4301. Hereinafter, with reference to FIG. 47, an operation when the pause control unit 4314 receives a pause release request from the user 4303 will be described.

  In FIG. 47, reference numbers 4741 and 4742 indicate a time axis, and the corresponding positions in the vertical direction indicate the same time. Reference numeral 4733 indicates the pause release request received from the user 4303 by the pause control unit 4314 and the reception time thereof. Reference numbers 4706B and 4707B indicate a data acquisition request issued to the data storage device 4302 by the data acquisition unit 4311 and its issue time. Reference numbers 4714B and 4715B indicate two blocks that are still stored in the buffer 4312 in the pause process, and a transmission time to the user 4303. Reference numbers 4716B and 4717B indicate the data acquired by the data acquisition requests 4706B and 4707B and the time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4724 indicates the maximum time D from when the data acquisition request 4706B is issued until the corresponding data 4716B is stored in the buffer 4312. Reference numerals 4863 to 4766 denote time intervals at which the data sending unit 4310 sends data to the user 4303.

  The user 4303 issues a pause release request 4733 and designates the data transmission rate after the pause release. The data transmission rate is specified by a data transmission interval Tsd ′ that is an average time interval between blocks to be transmitted, as in the case of a data transmission request.

  When the pause control unit 4314 receives a pause release request 4733 from the user 4303, the pause control unit 4314 refers to the buffer management unit 4313 and acquires the number N of blocks that are stored in the buffer 4312 after the pause process.

  The time at which the pause control unit 4314 controls the data acquisition unit 4311 to immediately resume data acquisition in response to the pause release request 4733 is defined as the data acquisition restart time trd. Similarly, the transmission time of the next block to be transmitted among the blocks stored in the buffer 4312, that is, the data transmission time 4714B of FIG. 47 is defined as the data transmission resumption time tsd.

The pause control unit 4314 substitutes the data transmission restart time tsd, the number N of blocks in the buffer 4312, the data transmission interval Tsd ′ after the pause release, and the data acquisition time D into the following equation (15). The data acquisition restart time trd is obtained.
trd = tsd + N · Tsd′−D (15)

  The pause control unit 4314 compares the data acquisition resumption time trd obtained using the above equation (15) with the data transmission resumption time tsd, and when the data acquisition resumption time trd is equal to or greater than the data transmission resumption time tsd, The time obtained by equation (15) is determined as the data acquisition restart time trd. When the data acquisition resumption time trd is less than the data transmission resumption time tsd, the tenth embodiment is applied and the pause control unit 4314 adds the data acquisition resumption time trd and the buffer 4312 to the equation (14). The data transmission restart time tsd is obtained by substituting the number N of blocks, the data transmission interval Tsd ′ after the pause release, and the data acquisition time D.

  The pause control unit 4314, in order from the data acquisition request 4706B for acquiring the block 4716B next to the last block 4715B stored in the buffer, to the data acquisition unit 4311 at the determined data acquisition restart time trd. Control to issue a data acquisition request until the end block is reached. Similarly, the pause control unit 4314 transmits data to the data transmission unit 4310 at the data transmission restart time tsd immediately from the first block 4714B stored in the buffer 4312 until the end block is transmitted. Control as follows.

  As described above, according to the eleventh embodiment, when an arbitrary transmission rate is designated as the data transmission rate after the pause is released, the pause control unit 4314 resumes data acquisition using the above equation (15). By obtaining the time, the data can be transmitted at a designated transmission speed without interruption after the pause is released.

(Twelfth embodiment)
Hereinafter, a file system according to a twelfth embodiment of the present invention will be described with reference to the drawings. The twelfth embodiment differs from the tenth embodiment shown in FIG. 43 in the function of each block, but the overall configuration is the same. Accordingly, the twelfth embodiment will be described below with the aid of the configuration of the tenth embodiment shown in FIG.

  In the twelfth embodiment, the operation when the server apparatus 4301 receives a data transmission request from the user 4303 is the same as the operation described with reference to FIG. 44 in the tenth embodiment. Similarly, the operation when the pause control unit 4314 receives a pause request from the user 4303 is the same as the operation described with reference to FIG. 45 in the tenth embodiment. Therefore, description of these operations is omitted.

  FIG. 48 is a time chart for explaining a method of calculating the data transmission resumption time in the pause release operation. FIG. 49 is a time chart showing the relationship among the data transmission request, the data acquisition request, and the data transmission in the pause release operation of the server apparatus 4301. Hereinafter, with reference to FIG. 48 and FIG. 49, an operation when the pause control unit 4314 receives a pause release request from the user 4303 will be described.

  In FIG. 48, reference numerals 4841 and 4842 indicate a time axis, and the corresponding positions in the vertical direction indicate the same time. Reference numeral 4833 indicates the pause release request received from the user 4303 by the pause control unit 4314 and the reception time thereof. Reference numeral 4806C indicates a data acquisition request issued by the data acquisition unit 4311 to the data storage device 4302 and its scheduled issue time. Reference numbers 4814C and 4815C indicate the two blocks that are still stored in the buffer 4312 in the pause process, and the scheduled transmission time to the user 4303. Reference numeral 4816C indicates data scheduled to be acquired by the data acquisition request 4806C and a scheduled time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4825 indicates the maximum time D from when the data acquisition request 4806C is issued until the corresponding data 4816C is stored in the buffer 4312. Reference numbers 4867 and 4868 indicate time intervals at which the data sending unit 4310 sends data to the user 4303.

  In FIG. 49, reference numeral 4833 indicates a pause release request received from the user 4303 by the pause control unit 4314 and its reception time. Reference numbers 4805C ′, 4806C, and 4807C indicate the issue times of data acquisition requests issued to the data storage device 4302 by the data acquisition unit 4311. Reference numeral 4814C ′ indicates a block that remains stored in the buffer 4312 after the pause process, and a transmission time to the user 4303. Reference numbers 4815C ′ and 4816C indicate the data acquired by the data acquisition requests 4805C ′ and 4806C and the time when the data transmission unit 4310 transmits the data to the user 4303. Reference numeral 4826 indicates the maximum time D from when the data acquisition request 4805C ′ is issued until the corresponding data 4815C ′ is stored in the buffer 4312. Reference numbers 4869 to 4872 indicate time intervals at which the data sending unit 4310 sends data to the user 4303.

  The user 4303 issues a pause release request 4833 and designates the data transmission rate after the pause release. The data transmission rate is specified by a data transmission interval Tsd ′ that is an average time interval between blocks to be transmitted, as in the case of data transmission request.

  When the pause control unit 4314 receives a pause release request 4833 from the user 4303, the pause control unit 4314 refers to the buffer management unit 4313 and acquires the number N of blocks that are stored in the buffer 4312 after the pause process.

  In response to the pause release request 4833, a time at which the pause control unit 4314 controls the data acquisition unit 4311 to immediately resume data acquisition is defined as a data acquisition restart time trd. Similarly, among the blocks stored in the buffer 4312, the transmission time of the next block to be transmitted, that is, the data transmission time 4814C of FIG. 48 and the data transmission time 4814C ′ of FIG. Define.

  The pause control unit 4314 obtains the data transmission resuming scheduled time tsd using the above-described equation (14). Next, the pause control unit 4314 compares the data transmission resumption scheduled time tsd obtained using the equation (14) with the data acquisition resumption time trd, and the data transmission resuming scheduled time tsd is equal to or greater than the data acquisition resumption time trd. In this case, the time obtained by Expression (14) is determined as the data acquisition restart time tsd, and the following processing is performed in the same manner as the pause release processing described in the tenth embodiment. That is, the pause control unit 4314 finishes in order from the data acquisition request for acquiring the next block immediately after the last block stored in the buffer 4312 to the data acquisition unit 4311 at the data acquisition restart time trd. Control to issue a data acquisition request until the block is reached. Similarly, the pause control unit 4314 transmits data to the data transmission unit 4310 at the determined data transmission restart time tsd until the end block is transmitted in order from the first block stored in the buffer 4312. Control to do.

When the scheduled data transmission resumption time tsd obtained using the equation (14) is less than the data acquisition resumption time trd, that is, as shown in FIG. 48, the following processing is performed. That is, the pause control unit 4314 uses the following equation (16) to calculate the number L of blocks that can be transmitted at the data transmission rate Tsd ′ after the pause release, that is, the limit data amount at the data acquisition time 4825 in FIG. Ask. However, in the following equation (16), [X] is the maximum integer not exceeding X.
L = [D / Tsd '] (16)

Next, the pause control unit 4314 obtains the number of blocks Nd to be deleted from the buffer 4312 using the following equation (17), and the buffer management unit 4313 uses the following equation (17) from the blocks stored in the buffer 4312. Control is performed so as to delete the Nd blocks obtained in (1).
Nd = N−L (17)

  The buffer management unit 4313 deletes the Nd blocks from the buffer 4312 in the order in which they were last stored in the buffer, that is, in descending order of block numbers. As a result, the number of blocks in the buffer 4312 matches L. As a result, the block 4815C in FIG. 48 is deleted from the buffer 4312.

The pause control unit 4314 substitutes the data acquisition restart time trd, the number L of blocks in the buffer 4312, the data transmission interval Tsd ′ after the pause release, and the data acquisition time D into the following equation (18). Then, a new data transmission restart time tsd ′ is obtained.
tsd ′ = trd + DL · Tsd ′ (18)

  Next, the pause control unit 4314 sends a data acquisition request 4805C for acquiring the block 4815C ′ next to the last block 4814C ′ stored in the buffer 4312 to the data acquisition unit 4311 at the data acquisition restart time trd. Control to issue data acquisition requests in order from 'until the end block is reached. Similarly, at the determined data transmission restart time tsd ′, the pause control unit 4314 immediately transmits the end block to the data transmission unit 4310 in order from the first block 4814C ′ stored in the buffer 4312. Control to send data.

  As described above, according to the twelfth embodiment, when an arbitrary transmission rate is designated as the data transmission rate after the pause release, data exceeding the limit data amount is deleted from the buffer 4312. The server apparatus 4301 does not need to have a buffer capacity larger than the limit data amount, and can transmit data at a designated transmission rate without interruption of data after the pause is released.

  The file system according to the present invention is useful as a file system including a scheduling device that can provide data with a quick response time in response to a request from a user.

The block diagram which shows the structure of the file system which concerns on the 1st Embodiment of this invention The block diagram which shows the structure of the part relevant to the file access in each server unit of FIG. The block diagram which shows the structure of the part relevant to the file access in the file manager of FIG. The figure which showed the relationship between resource reservation and allocation, and data read-out operation | movement in 1st Embodiment. The figure which shows an example of the MS arrival slot management table which manages the state of MS arrival slot The figure which shows an example of the slot management table for writing which manages the state of the slot for writing The figure which shows an example of a stream management table The block diagram which shows the more detailed structure of the resource management part shown in FIG. The block diagram which shows the structure of the file system which concerns on the 2nd Embodiment of this invention. The block diagram which shows the structure of the file access part in each I / O unit of FIG. The block diagram which shows the structure of the file access part in each block storage unit of FIG. The block diagram which shows the structure of the file system which concerns on the 3rd Embodiment of this invention. A block diagram showing a configuration of a disk management apparatus according to a fourth embodiment of the present invention. The figure which shows the bandwidth priority layout which determines the ratio which distributes the data in proportion to the bandwidth of the storage device A diagram showing a capacity-priority layout that determines the ratio of data distribution in proportion to the capacity of the storage device The figure which shows the band setting layout which decides the ratio which distributes the data by mixing the band priority layout and the capacity priority layout FIG. 13 is a block diagram showing the configuration of a file system using the disk management apparatus shown in FIG. The block diagram which shows the structure of the scheduling apparatus which concerns on the 6th Embodiment of this invention. The block diagram which shows the structure of the video information provision system using the scheduling apparatus of FIG. 19 is a sequence diagram showing how signals are transmitted and received between the video information providing apparatus 1920 and the clients (1 to n) 1921 in FIG. The figure which shows the registration content of the execution right generation cycle table 1812 of FIG. The figure which shows the queuing content of the process waiting queue 1814 of FIG. The figure which shows the queuing content of the execution right queue 1816 of FIG. Flowchart for explaining the operation in which the execution right generation unit 1813 in FIG. 18 generates the execution right. 18 is a flowchart for explaining an operation in which the execution instruction unit 1815 in FIG. 18 instructs execution of a read request. The block diagram which shows the structure of the scheduling apparatus which concerns on the 7th Embodiment of this invention. The block diagram which shows the structure of the video information provision system using the scheduling apparatus of FIG. The figure which shows the registration content of the execution right generation cycle table 2692 of FIG. The figure which shows the queuing content of the process waiting queue 2694 of FIG. The figure which shows the queuing content of the execution right queue 2696 of FIG. Diagram for explaining the operation of another execution right queue The block diagram which shows the structure of the data transmission system using the server apparatus which concerns on the 8th Embodiment of this invention. FIG. 32 is a block diagram showing a more detailed configuration of the I / O unit 3207 in FIG. The flowchart which shows the operation | movement of the command reception part 3327 in FIG. The flowchart which shows operation | movement of the command queue control part 3322 in FIG. The flowchart which shows operation | movement of the command transmission part 3323 in FIG. The block diagram which shows the structure of the data transmission system using the server apparatus concerning the 9th Embodiment of this invention. FIG. 37 is a block diagram showing a more detailed configuration of the I / O unit 3707 in FIG. 37 is a flowchart showing the operation of the resource management unit 3701 in FIG. 38 is a flowchart showing the operation of the command receiving unit 3847 in FIG. 38 is a flowchart showing the operation of the command queue control unit 3842 in FIG. 38 is a flowchart showing the operation of the command sending unit 3843 in FIG. The block diagram which shows the structure of the file system which concerns on the 10th Embodiment of this invention. In the tenth embodiment, a time chart showing a relationship among a data transmission request, a data acquisition request, and data transmission in a data transmission operation of the server device In the tenth embodiment, a time chart showing a relationship among a data transmission request, a data acquisition request, and data transmission in the pause operation of the server device In the tenth embodiment, a time chart showing a relationship among a data transmission request, a data acquisition request, and data transmission in the pause release operation of the server device In the eleventh embodiment, a time chart showing a relationship among a data transmission request, a data acquisition request, and data transmission in the pause release operation of the server device In the twelfth embodiment, a time chart for explaining a method for calculating a data transmission restart time in a pause release operation of a server device In the twelfth embodiment, a time chart showing a relationship among a data transmission request, a data acquisition request, and data transmission in the pause release operation of the server device

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Switching apparatus 2 File manager 3 Server unit 101 External message interface part 102 Resource assignment part 103 Slot assignment part 104 Address search part 105 Request sending part 106 External network I / O control part 107 Internal message interface part 108 Storage apparatus 109 Storage control part 110 Internal network I / O control unit 111 Report value management unit 201 Message interface unit 202 Resource management unit 401 Report parameter processing unit 402 Write bandwidth calculation unit 403 Acceptance determination unit 404 Resource storage unit 901 File manager 902 I / O unit 905 Block storage Unit 908 Switching apparatus 1001 External message interface unit 1002 Resource allocation unit 1003 Slot allocation unit 1004 Address search unit 1005 Request sending unit 1006 External network I / O control unit 1007 Internal message interface unit 1008 Declared value management unit 1101 Internal message interface unit 1102 Internal network I / O control unit 1103 Storage control unit 1104 Storage device 1201 Message interface unit 1202 Resource management unit 1203 Read / write control unit 1204 Storage device 1205 Declared value management unit 1300 Disk management device 1301 Disk bandwidth management unit 1302 Disk capacity management unit 1303 Distribution ratio determination unit 1304 Bandwidth calculation unit 1305 Capacity calculation unit 1306 Total disk bandwidth control unit 1307 Total disk capacity control units 1708a and 1708b Storage device 1709 File attribute management unit 1710 Band management unit 1711 File open Processing unit 1712 Band acquisition unit 1713 Input / output control units 1414a to 1414c Storage devices 1515a to 1515c Storage devices 1616a to 1616c Storage devices 1810 and 2690 Scheduling devices 1811 and 2691 Registration instruction units 1812 and 2692 Execution right generation cycle tables 1813 and 2693 Execution rights Generation unit 1814, 2694 Processing queue 1815, 2695 Execution instruction unit 1816, 2696 Execution right queue 1920, 2700 Video information providing device 1921, 2701 Client 1922, 2702 Server 2703 Transmission path 3201, 3701 ... Resource management unit 3202, 3702 ... Data Storage units 3203, 3703 ... exchange units 3204, 3704 ... input / output units 3205, 3705 ... storage devices 3206, 3706 ... MSFS
3207, 3707 ... I / O units 3208, 3708 ... user terminals 3301, 3801 ... priority input / output control units 3302, 3802 ... non-priority input / output control units 3321, 3841 ... command queues 3322, 3842 command queue control units 3323, 3843 ... Command sending unit 3324, 3844 ... Data buffer unit 3325 ... Buffer management unit 3326, 3848 ... Data sending unit 3327, 3847 ... Command receiving unit 3848 ... Counter 3849 ... Sending queue 4301 Server device 4302 Data storage device 4310 Data sending unit 4311 Data acquisition Unit 4312 buffer 4313 buffer management unit 4314 pause control unit

Claims (42)

A scheduling device that schedules a service request so that a service is periodically provided to a plurality of clients from a server,
First queuing means for queuing service requests from the plurality of clients;
Registration means for registering a cycle for generating an execution right for each of the plurality of clients;
Execution right generation means for generating an execution right for executing a service request in relation to the period registered in the registration means;
Execution instruction means for instructing the server to execute a service request from a client corresponding to the execution right generated by the execution right generation means among the service requests queued in the first queue means; A scheduling apparatus comprising:   The execution right generation means is activated at a predetermined cycle, and at the time of activation, the execution right generation unit generates an execution right based on a cycle in which the execution right is registered and a cycle in which the execution right is activated. 2. The scheduling apparatus according to claim 1, wherein, for each client, a determination is made for each client, and the execution rights of the clients determined to generate the execution right are generated in a lump.   The execution right generation means generates a next execution right from the current activation time based on the period in which the execution right should be generated and the period in which the execution right is registered, registered in the registration means. The time until the time is calculated, and it is determined whether or not the execution right is generated by comparing the time obtained by the calculation and the cycle in which the self is activated. Scheduling device.   The execution right generation means generates an execution right for a client whose time Tofs from the current time when it is activated to the time when the next execution right is generated is Tofs <Tslt with respect to the cycle Tslt in which the self is activated. Furthermore, the Tofs of the client is added with the period Treq that is registered in the registration means and is to be generated, and when the Tofs obtained by the addition is Tofs ≧ Tslt, the Tofs The scheduling apparatus according to claim 3, wherein a new Tof is calculated by the expression Tofs = Tofs−Tslt and is set as Tofs for the next activation.   When the Tofs obtained by the addition is Tofs <Tslt, the execution right generating means further repeats the operation of generating the execution right and adding Treq to the Tofs and comparing it with Tslt. When the Tofs obtained in this way becomes Tofs ≧ Tslt, a new Tofs is calculated with respect to the Tofs by the formula Tofs = Tofs−Tslt, and the Tofs for the next activation is obtained. The scheduling apparatus according to claim 4.   The execution right generation means has a time Tofs from the current activation time to a time to generate the next execution right, Tofs ≧ Tslt, or a service request is queued in the first queue means For a client that has not been executed, an execution right is not generated, and a new Tofs is calculated for the Tofs by the formula Tofs = Tofs−Tslt, and the Tofs for the next activation is obtained. The scheduling apparatus according to claim 4.   7. The scheduling according to claim 6, wherein when the new Tofs obtained by calculation is Tofs <0, the execution right generation unit sets the Tofs at the next activation to 0. 8. apparatus.   The execution instruction means is activated at a predetermined cycle, and selects an execution right of a client having a shorter time from the current activation time to a time when the next execution right is generated. Item 3. The scheduling device according to Item 2. The execution right generation means calculates an evaluation index indicating a priority as a function of a cycle in which the execution right is to be registered, and is given to the generated execution right.
The execution instruction means selects an execution right to be prioritized from the execution rights generated by the execution right generation means based on the assigned evaluation index. Scheduling device. The execution right generation means uses the function Nsft = f (Treq / Tslt) for the evaluation index Nsft with respect to the period Treq for generating the execution right registered in the registration means and the period Tslt at which the execution right is activated. Where f (x) is the smallest integer greater than or equal to x,
The scheduling apparatus according to claim 9, wherein the execution instruction unit selects an execution right having a smaller evaluation index Nsft.   The scheduling apparatus according to claim 9, wherein the execution right generation unit is activated at a cycle shorter than the longest cycle registered in the registration unit.   The scheduling apparatus according to claim 10, wherein the execution right generation unit is activated at a cycle shorter than the shortest of the cycles registered in the registration unit.   The execution instructing means includes second queuing means for queuing the execution rights generated by the execution right generation means for each given evaluation index and in the order in which they are generated. The scheduling apparatus according to claim 9, wherein, among the execution rights that have been assigned, the one that has been assigned with a smaller evaluation index and that has been generated earlier is selected.   The execution instructing unit recognizes the maximum number of service requests that are instructed to be executed at the time of activation, and stops instructing execution when the number of service requests to be executed is equal to the maximum number. The scheduling apparatus according to claim 9, wherein the evaluation index of the executed right is changed to one having a higher priority.   The maximum number of service requests that are recognized by the execution instructing unit and that are instructed to execute at the time of start-up are the maximum number of service requests that can be executed by the server within a time period equal to a period in which the server is started. The scheduling apparatus according to claim 14.   The execution instructing unit recognizes the maximum number of service requests that are instructed to be executed at the time of activation, and stops instructing execution when the number of service requests to be executed is equal to the maximum number. The scheduling apparatus according to claim 10, wherein the execution index is decremented.   The maximum number of service requests that are recognized by the execution instructing unit and that are instructed to execute at the time of start-up are the maximum number of service requests that can be executed by the server within a time period equal to a period in which the server is started. The scheduling apparatus according to claim 16.   The scheduling apparatus according to claim 14, further comprising a registration instruction unit that instructs the registration unit to register a cycle in which an execution right should be generated based on reports from the plurality of clients.   The registration instruction means recognizes the maximum number of service requests to be executed by the execution instruction means per unit time, and is notified to register another period in addition to the already registered periods. If it is determined that the average number of execution rights generated per unit time does not exceed the maximum number even if further registration is performed, the registration means is instructed to register the cycle, and further registration is performed. The scheduling apparatus according to claim 18, wherein, if it is determined that the number is exceeded, registration is not performed. A scheduling device that schedules service requests so that services are periodically provided to clients from a plurality of servers,
First queuing means for queuing service requests to the plurality of servers;
Registration means for registering a cycle for generating an execution right for each of the plurality of servers;
Execution right generation means for generating an execution right for executing a service request in relation to the period registered in the registration means;
Execution instruction means for instructing the server to execute a service request to the server corresponding to the execution right generated by the execution right generation means among the service requests queued in the first queue means; A scheduling apparatus comprising:   The execution right generation means is activated at a predetermined cycle, and at the time of activation, the execution right generation unit generates an execution right based on a cycle in which the execution right is registered and a cycle in which the execution right is activated. The scheduling apparatus according to claim 20, wherein the server execution right is determined for each server and the execution right of the server determined to generate the execution right is generated in a lump.   The execution right generation means generates a next execution right from the current activation time based on the period in which the execution right should be generated and the period in which the execution right is registered, registered in the registration means. 23. The method according to claim 21, wherein a determination is made as to whether or not to generate an execution right by calculating a time until the execution right and comparing a time obtained by the calculation with a cycle in which the self is activated. Scheduling device.   The execution right generation means generates an execution right of a server whose time Tofs from the current time when it is activated to the time when the next execution right is generated is Tofs <Tslt with respect to the cycle Tslt in which it is activated. Furthermore, the period Ttv for generating the execution right registered in the registration unit is added to the Tofs of the server, and when the Tofs obtained by the addition is Tofs ≧ Tslt, the Tofs 23. The scheduling apparatus according to claim 22, wherein a new Tof is calculated by the expression Tofs = Tofs-Tslt and is set as Tofs for the next activation.   When the Tofs obtained by the addition is Tofs <Tslt, the execution right generation means further repeats the operation of generating an execution right and adding Titv to the Tofs and comparing it with Tslt. When the Tofs obtained in this way becomes Tofs ≧ Tslt, a new Tofs is calculated with respect to the Tofs by the formula Tofs = Tofs−Tslt, and the Tofs for the next activation is obtained. The scheduling apparatus according to claim 23.   The execution right generation means has a time Tofs from the current activation time to a time to generate the next execution right, Tofs ≧ Tslt, or a service request is queued in the first queue means For a server that has not been executed, an execution right is not generated, and a new Tofs is calculated with respect to the Tofs by the formula Tofs = Tofs-Tslt, and the Tofs for the next startup is obtained. The scheduling apparatus according to claim 23.   26. The scheduling according to claim 25, wherein when the new Tofs obtained by calculation is Tofs <0, the execution right generation unit sets the Tofs at the next activation to 0. apparatus.   The execution instruction means is activated at a predetermined cycle, and selects an execution right of a server having a shorter time from the current activation time to a time when the next execution right is generated. Item 22. The scheduling device according to Item 21. The execution right generation means calculates an evaluation index indicating a priority as a function of a cycle in which the execution right is to be registered, and is given to the generated execution right.
28. The execution instruction unit according to claim 27, wherein the execution instruction unit selects an execution right to be given priority among execution rights generated by the execution right generation unit based on the assigned evaluation index. Scheduling device. The execution right generation means uses the function Nsft = f (Titv / Tslt) for the evaluation index Nsft with respect to the period Ttv for generating the execution right registered in the registration means and the period Tslt at which the execution right is activated. Where f (x) is the smallest integer greater than or equal to x,
The scheduling apparatus according to claim 28, wherein the execution instruction means selects an execution right having a smaller evaluation index Nsft.   30. The scheduling apparatus according to claim 29, wherein the execution right generation unit is activated at a cycle shorter than the longest cycle registered in the registration unit.   30. The scheduling apparatus according to claim 29, wherein the execution right generation unit is activated at a cycle shorter than a shortest cycle registered in the registration unit.   The execution instructing means includes second queuing means for queuing the execution rights generated by the execution right generation means for each given evaluation index and in the order in which they are generated. 29. The scheduling apparatus according to claim 28, wherein, among the execution rights that have been assigned, the one with the smaller evaluation index assigned and that has been generated earlier is selected.   The execution instructing unit recognizes the maximum number of service requests that are instructed to be executed at the time of activation, and stops instructing execution when the number of service requests to be executed is equal to the maximum number. 29. The scheduling apparatus according to claim 28, wherein the evaluation index of the executed right is changed to one having a higher priority.   The maximum number of service requests that are recognized by the execution instructing unit and that are instructed to execute at the time of activation corresponds to the maximum amount of information that can be transmitted from all servers to the client within a time equal to the period in which the self is activated. The scheduling apparatus according to claim 33, wherein the scheduling apparatus is the number of service requests.   The execution instructing unit recognizes the maximum number of service requests that are instructed to be executed at the time of activation, and stops instructing execution when the number of service requests to be executed is equal to the maximum number. 30. The scheduling apparatus according to claim 29, wherein the execution right evaluation index is decremented.   The maximum number of service requests that are recognized by the execution instructing unit and that are instructed to execute at the time of activation corresponds to the maximum amount of information that can be transmitted from all servers to the client within a time equal to the period in which the self is activated. 36. The scheduling apparatus according to claim 35, wherein the scheduling apparatus is the number of service requests.   34. The scheduling apparatus according to claim 33, further comprising registration instruction means for instructing the registration means to register / update a cycle for generating an execution right based on a report from the client.   The registration instructing means recognizes the maximum number of service requests that the execution instructing means instructs all servers to execute per unit time, and in addition to the already registered periods, the period of other servers When it is determined that the average number per unit time of the execution rights of all servers generated even if further registration is made when it is declared to be registered, the registration means is instructed. 38. The scheduling apparatus according to claim 37, wherein the period is registered, and if it is further registered, it is not registered when it is determined that the period is exceeded.   The registration instructing unit recognizes the maximum number of service requests that the execution instructing unit instructs each server to execute per unit time, and in addition to the already registered cycle, the cycle of another server is further determined. The service request in which the execution instruction means instructs the server to execute the unit per unit time, the average number of execution rights of the server that is generated even when the registration is declared. If it is determined that the maximum number is not exceeded, the registration unit is instructed to register the period, and if it is further registered, if it is determined that the number is exceeded, registration is not performed. The scheduling device described.   The maximum number of service requests recognized by the registration instructing unit per unit time that the execution instructing unit instructs each server to execute is the maximum number of service requests that can be executed by each server per unit time. 40. The scheduling apparatus according to claim 39, wherein there is a scheduling apparatus.   The registration instruction means includes a maximum number of service requests that the execution instruction means instructs all servers to execute per unit time, and a maximum number of service requests that the execution instruction means instructs each server to execute per unit time. When the report is received to update the already registered period Ttv for generating the execution right of a server to the one that generates the execution right by adding the period Treq, Even if it is updated, it is determined that the average number of execution rights of all servers per unit time generated does not exceed the maximum number of service requests that the execution instruction unit instructs all servers to execute per unit time. In addition, the average number of execution rights of the server generated per unit time is a server that the execution instruction means instructs the server to execute per unit time. If it is determined that the maximum number of requests is not exceeded, the registration means is instructed to update the already registered Titv to a new Titv calculated by the formula Titv = 1 / (1 / Titv + 1 / Treq). 38. The scheduling apparatus according to claim 37, wherein, if it is updated, if it is determined that it is exceeded, it is not updated.   When the registration instructing unit receives a declaration to update the already-registered period Titv for generating the execution right of a server to one that generates the execution right excluding the period Treq, the registration instructing unit The scheduling according to claim 37, wherein the previously registered Titv is updated to a new Titv calculated by the expression Titv = 1 / (1 / Titv-1 / Treq). apparatus.

Images