JP2008158844A - Resource management device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To achieve the high efficiency of the management function of network resources or computer resources. <P>SOLUTION: This resource management device performs assignment of network resources or computer resources necessary for execution of a job by identifying a type of the job, and to refer to a job database showing a service class corresponding to the type of the job. Thus, it is possible for a user device to receive the provision of resources with quality corresponding to the service of the job only by transmitting job identification information. <P>COPYRIGHT: (C)2008,JPO&INPIT

Description

  The present invention relates to a technique for providing scheduling efficiency of a resource management system that provides network resources or computer resources to a user.

  In the present invention, the computer resource is a resource having a function of executing a job such as a computer device or a server connected to the network, and the network resource is a transmission path on the network or a path set in this transmission path. And a node device including a node device having a function of setting or managing the path, and a resource for connecting computer resources. In addition, simply speaking, “resources” refers to computer resources and / or network resources.

  A distributed computing system (distributed system) includes various resources such as a computer, a storage device, and a network. For this reason, in order to obtain a quality of service (QoS) required for a certain job (job), a resource management technique that guarantees the service quality over a plurality of resources is required.

  In particular, in grid computing in which resources are shared among a plurality of domains managed independently, it is often not permitted to centrally manage the resources of the entire system. This type of system requires a QoS guarantee function using a plurality of independent resource management mechanisms.

  As a conventional function for guaranteeing QoS in a distributed system, advance reservation of resources is known. Advance reservation is an operation for guaranteeing QoS in a certain period of time for each resource necessary for job execution.

  1 and 2 show a conventional resource management system (see, for example, Patent Document 1). FIG. 1 is an overall configuration diagram of a conventional resource management system. FIG. 2 is a sequence diagram showing a communication procedure between a conventional resource management device and a user device. The user device 1 makes a reservation request to the resource management device 2 (S1). If the reservation is available, a reservation certificate is issued to the user device 1 (S2). When the reserved time is reached, the resource management device 2 provides the computer resource 3 and the network resource 4 to the user device 1 (S3), so that the user device can use the computer resource 3 via the network resource 4. .

  In addition, an experiment of a grid service NW (network) in which the user device 1 inputs the job start time and end time to the grid scheduler so that the resource scheduler 2 reserves the resource and executes the job is performed. (For example, refer nonpatent literature 1).

JP 2000-259537 A T.A. Kudoh et al, "G-lamda: an interface for bandwidth reservation from applications and middleware" AGNM 2006

  In the conventional technology, the user device can reserve the necessary amount of resources when necessary, but the user device cannot predict the job execution time, for example. When the time is estimated to be more than the actual execution time, the user device can only make a reservation when the user device releases the reservation when the time is over. Also, if the time is estimated to be shorter than the actual execution time, the current technology may discard the result calculated so far.

  In order to solve such a problem, it is only necessary to accurately calculate the necessary resources in the user device and to reserve the resource without excess or deficiency, but it is desirable to ask the user device for such a burden. Absent. In addition, since it is difficult for the user device to grasp all the resources of the entire network, it is difficult to obtain resource information that is the basis of the calculation even if the user device tries to accurately calculate the resource that is reserved. .

  Therefore, it is useful to have a function of managing resource reservations from user devices on the network side and perform efficient resource allocation.

  The present invention has been made under such a background, and it is an object of the present invention to provide a resource management apparatus and method capable of realizing a highly efficient management function of network resources and computer resources. .

  As described above, for example, identifying the type of job or predicting the execution time of the job is effective for maximizing the use of resources. Therefore, the present invention is characterized in that the type of job is identified and optimal resource allocation is performed. This makes it possible, for example, to distinguish between jobs that can be executed in a divided manner and those that require continuous reservations, and for those that can be executed in a divided manner, reservations are made for reservation time slots that are vacant. This makes it possible to effectively use resources. Hereinafter, a technique for allocating resources according to the type of job, the estimated execution time, and the required reliability will be described.

  Further, in the present invention, the resource management device refers to the job database indicating the service class corresponding to the job type, so that the user device can provide quality resources according to the job service simply by sending job identification information. Can receive.

  That is, the present invention receives a resource reservation request transmitted from a user device, determines whether or not the corresponding resource is free, and if there is a free resource, allocates a resource that satisfies the resource reservation request to the user device. It is a management device.

  Here, the present invention is characterized by a job database in which correspondence information between identification information indicating the type of job and service class information necessary for each job is recorded, network resources, and quality of the network resources. Resource information holding means for holding the service class information shown, and when a resource reservation request including job identification information is received from the user apparatus, the job database is referred to and the service class information of the job is searched from the job identification information And a means for determining a network resource satisfying the required quality of the retrieved service class based on service class information indicating the quality of the network resource held in the resource information holding means and allocating it to the user apparatus. .

  In this way, by referring to the job database, it is possible to provide resources of optimum quality to the user and satisfy the user's QoS.

  For example, the network resource includes a transmission path on the network or a path set in the transmission path, and the service class information includes information on a delay amount or packet loss tolerance of the path or transmission path. Thus, the user's QoS can be satisfied by including information on the delay amount of the network resource or the allowable packet loss amount in the job database.

  Further, the job database holds a correspondence relationship between a job process and a history of past execution times for each process of the job, and upon receiving a resource reservation request from a user device, the job database is referred to and the resource reservation request is referred to. Based on the identification information of the job included in the job, the correspondence relationship between the job and its execution time history is referred to, the execution time of the job is predicted, and a resource satisfying the predicted execution time is determined and allocated to the user device. Means may be provided.

  In this way, by referring to the job database, it is possible to know in advance the time required to execute a job, and it is possible to improve resource efficiency.

  Further, the means for comparing the service class of the network resource connected to the user device and the service class of the job being executed or scheduled to be executed by the user device and the user device are connected based on the comparison result by the means for comparing. If the network resource cannot satisfy the required quality of the job being executed or scheduled to be executed by the user device, means for searching for a computer resource having the network resource that satisfies the required quality of the user device; and When a computer resource having a network resource that satisfies the required quality of the user device is found based on a search result, a network resource that becomes a connection path between the user device and the computer resource is allocated to the user device, and the execution or execution is being performed. Schedule the job to the computer resource It can comprise means to move to.

  In this way, even if the network resource connected to the user device cannot execute the job of the user device satisfactorily, the network resource and the computer resource that can execute the job of the user device satisfactorily are searched for. By using the computer resource as a proxy and allocating a network resource as a connection path between the user device and the computer resource to the user device, it is possible to efficiently provide a resource that satisfies the required quality of the user device.

  A calculation result information collecting unit for collecting calculation result information from the user device; a unit for determining a job execution result based on the calculation result collected by the calculation result information collecting unit; and a determination result by the determining unit. If the calculation result is found to be abnormal based on the calculation result information discarding and prompting the user device to recalculate, and the execution time of the job in the calculation result information collecting process by the calculation result information collecting means Means for extending the reservation time or re-reserving when the time exceeds the limit.

  Thus, by quickly discarding invalid calculation results, it is possible to improve the efficiency of job execution and improve the reliability of the calculation results. Furthermore, it can be avoided that the reservation time ends in the middle of the job and the job becomes invalid.

  In addition, upon receiving a reliability improvement request from a user device, a means for allocating a plurality of redundant network resources satisfying the required quality for one job request, and bypassing the failed network resource when a network resource failure occurs Means for switching to network resources. Thereby, the reliability of a network can be improved.

  In addition, upon receiving a request for improving reliability from a user apparatus, means for allocating a plurality of redundant computer resources for one job request, and a computer in which a failure has not occurred in a computer resource in which a failure has occurred when a computer resource failure has occurred Means for switching to resources. Thereby, the reliability of computer resources can be improved.

  The present invention can also be viewed from a viewpoint as a resource management method. That is, the present invention receives a resource reservation request transmitted from a user device, determines whether or not the corresponding resource can be allocated to the user, and if the resource can be allocated, assigns the resource satisfying the resource reservation request to the user. This is a resource management method executed by a resource management device assigned to a device.

  Here, a feature of the present invention is that the resource management device records a correspondence relationship between identification information indicating a job type and service class information necessary for each job in a job database, and the network resource and the network Service class information indicating the quality of the resource is held in the resource information holding means, and when a resource reservation request including job identification information is received from the user apparatus, the job database is referred to and the job class of the job is determined from the job identification information. The resource satisfying the required quality of the retrieved service class is determined based on the service class information indicating the quality of the resource held in the resource information holding means and allocated to the user apparatus.

  The present invention can also be viewed from the viewpoint of a program. That is, the present invention is a program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize a function corresponding to the function of the resource management apparatus of the present invention.

  By recording the program of the present invention on a recording medium, the general-purpose information processing apparatus can install the program of the present invention using this recording medium. Alternatively, the program of the present invention can be directly installed on the general-purpose information processing apparatus via a network from a server that holds the program of the present invention.

  Thereby, the function corresponding to the resource management apparatus of this invention is realizable using a general purpose information processing apparatus.

  The program of the present invention includes not only a program that can be directly executed by a general-purpose information processing apparatus but also a program that can be executed by installing it on a hard disk or the like. Also included are those that are compressed or encrypted.

  According to the present invention, it is possible to achieve high efficiency of management functions for network resources and computer resources.

  The resource management apparatus according to the present invention can provide a technology for effectively utilizing resources by predicting the time required for job execution and providing resources of quality according to the service. Hereinafter, this resource management apparatus will be described in detail.

(First embodiment)
A first embodiment will be described with reference to FIGS. FIG. 3 is a functional block diagram of the resource management apparatus in the first embodiment. FIG. 4 is an overall configuration diagram of the resource management system for explaining the first embodiment. FIG. 5 is a sequence diagram for explaining the first embodiment.

In this embodiment, the network is MPLS (Multi Protocol Label).
Switching) or GMPLS (Generalized MPLS) is an example of providing path information. In addition, a configuration example including the function of the scheduler unit 22 for reserving resources in the resource management device 2 is shown. However, since the scheduler unit 22 may be outside the resource management device 2 as described in Non-Patent Document 1, it is not an essential element of the resource management device itself.

  The user device 1 sends a resource reservation request including the job start time and end time, job identification information, and network bandwidth to the scheduler unit 22 of the resource management device 2 (S10). The transmission of the reservation request may be instructed by the user of the user device 1 directly to the user device 1 or, when the user device 1 executes a job, the resource management device 2 automatically reserves necessary resources. It may be.

  The resource management device 2 sends the job identification information received from the user device 1 to the job database 20, and searches for and acquires information on the required service class for the requested job (S11, S12).

  Job service class information is determined not only by delay amount and packet loss tolerance, but also by delay variation and dispersion amount. For example, video distribution requires that real-time performance and packet loss be minimized. The reservation is made at # 1, and the calculation process is reserved at service class # 5, which is less restrictive than service class # 1 because the demand for delay and packet loss is lighter than video distribution.

  The resource management device 2 that has acquired the job service class information searches the route database 24 for network resources that satisfy the service class information (S13), and displays the network resources on the user device 1 (S14). If the user device 1 permits the network resource (S15), the resource management device 2 transmits a reservation certificate for certifying the provision (S16). The resource management device 2 provides network resources satisfying the service class from the network control unit 23 to the user device 1 (S17). In this way, the resource management device 2 can provide the user device 1 with service class resources suitable for the job.

  FIG. 6 shows an example of the job database 20. The job database 20 corresponds to job names and service class information (delay, packet loss, service class identification information (service class ID)). To update the information in the job database 20, the network operator updates the information, stores a correspondence table between jobs and service classes provided in the past, has a communication function with the job database of other resource management devices, and stores information. By synchronizing, the contents of the job database 20 can always be updated.

  Here, the user device 1 directly reserves the resource in the resource management device 2, but it may be via the scheduler unit 22. Also, a plurality of job identifiers can be reserved simultaneously, for example, requesting video distribution and calculation simultaneously.

  FIG. 7 shows an example of the route database 24. The job database 20 includes path identification information (path ID) that is a network resource, a service class corresponding to the path ID, and path information corresponding to the path ID.

  Note that the route database 24 holds only data link quality information between nodes even if there is no route information. When a reservation request is received, the network controller 23 calculates the route based on the reservation request to satisfy the request. May indicate a route.

(Second embodiment)
A second embodiment will be described with reference to FIGS. Grid computing systems may be used for various purposes such as physical calculation and finance, but many functions in programs used in jobs in the same field are the same, and if the job contents such as functions are known, jobs The calculation time required for execution can be estimated. For example, in an asset risk calculation performed by a financial institution, an approximate solution is obtained by repeatedly performing calculations using random numbers using the Monte Carlo method.

  FIG. 8 shows a functional block diagram of the resource management device 2 of the second embodiment. FIG. 9 shows a sequence diagram for explaining the second embodiment. FIG. 10 shows an example of the job database 20 used in the second embodiment. The overall configuration of the resource management system is the same as that shown in FIG. The job database 20 used in the second embodiment includes, in addition to the information of the job database 20 described in the first embodiment, functions, processes, subroutines, and the like of jobs executed in the past as shown in FIG. And a correspondence table of the time (time / number of times (seconds): that is, the execution time per one time) required for the execution. In the example of FIG. 10, the number of repetitions, accuracy, and number of resources are also recorded.

  The scheduler unit 22 of the resource management device 2 receives from the user device 1 a resource reservation request including information including the job contents described below (S21). Then, necessary parameters are sent to the execution time calculation unit 25, and the job execution time is predicted from jobs executed in the past by referring to the execution time calculation unit 25 (S22, S23, S24).

  The resource management device 2 searches the scheduler unit 22 for a resource that satisfies a condition based on the job execution time obtained from the job database 20 and the service class of the job (S25), and searches the search result (reservable resource list) for the user. The device 1 is notified (S26). If the user device 1 approves this resource (S27), the resource is provided to the user device 1, and a reservation certificate for certifying the provision is transmitted (S28).

  FIG. 10 shows an example of the job database 20. The job database 20 may have only a correspondence table as described above, or may have a past history. If the time of the job process is known, the job execution time can be predicted by calculating (T = F (Y) × X) as shown in the upper right. Here, T is the job execution time, F is the time required for processing the process (job function, etc.), X is the number of repetitions, and Y is the number of resources.

  There are two methods for knowing the functions, processes, subroutines, and the like included in the job, and a method in which the user directly inputs, and a method in which the resource management device 2 reads the contents of the job and extracts the functions. As a result, the time required for job execution can be predicted, and resource efficiency can be improved.

The reservation request of the user device 1 has the following parameters.
1. When did you run the job in the past? As shown in FIG. 10, the calculation can be performed by referring to the time taken for execution from the past history.
2. Functions, processes, subroutines, and the number of repetitions included in the job. This is a parameter for job execution time calculation.
3. Accuracy of job execution time calculation. Check how accurately the estimated job execution time needs to be calculated. As a result, the job execution time can be predicted efficiently. Of course, in order to increase the accuracy, it is necessary to input the input function in detail.
4). The job execution time desired by the user. The resource management device 2 divides and executes the job for each process so that the time desired by the user is reached, so that the user does not need to divide and input the job and the user does not need to be aware of the program. The job can be divided without ending the job halfway by knowing the subroutine and process time from the job database.

  When making a reservation, depending on the reliability of the estimated time, there is a method of taking a protection time longer than the estimated time in order to avoid completion of the calculation in the middle.

(Third embodiment)
In the case of large-capacity video distribution or the like, when a job is executed from the user device 1, the user device 1 also needs to have sufficient resources for video distribution.

  FIG. 11 is a diagram for explaining the third embodiment. Information on network resources such as network bandwidth and quality of the user apparatus 1 together with job identification information and job capacity is entered in the resource reservation request of the user apparatus 1 and sent to the scheduler unit 22 of the resource management apparatus 2. The resource management device 2 compares the service class of the job with the resources of the user device 1, and if the resource quality of the user device 1 is low, the resource management device 2 sends a job data transfer request to the user device 1. put out. For example, in FIG. 11, the job requires a bandwidth of 10 Gbps (thick solid line), whereas the network on the user side is 1 Gbps (thin solid line).

  Also, the resource management device 2 sends a message (CRM (Cluster Resource Management)) for inquiring whether there is a resource capable of providing a sufficient network provided by the job service class and the job service class. Send to each computer resource. As a result, it is possible to find a computer resource that has free capacity and can provide a resource of sufficient quality.

  Upon receiving the job data transfer request, the user device 1 transfers the job data to the free computer resource 3. At this time, since the distribution is not real-time, the network quality may be poor. The resource management device 2 can provide a high-quality service without depending on the resource of the user device 1 by providing the network resource 4 satisfying the service class to the migrated computer resource 3.

  Further, by allocating a network resource serving as a connection path to the computer resource 3 that executes the job on behalf of the user device 1 to the user device 1, the user device 1 can transfer the job to the computer resource 3, The execution result of the job completed by the computer resource 3 can be received. For such job migration, execution status monitoring, and job execution result reception, there is no problem even with network resources having a low service class.

  This is not limited to video distribution. If sufficient resources for job execution cannot be provided from the job database 20, the job is transferred to a computer resource having free capacity, and the network resource of the free resource and the job execution resource is connected. Thus, resources according to the service class can be provided.

(Fourth embodiment)
A fourth embodiment will be described with reference to FIGS. In the fourth embodiment, when calculation is performed by grid computing, a job is simultaneously executed with a plurality of resources, so that it is difficult for the resource management device 2 to manage each resource. It is considered difficult to guarantee.

  In addition, an erroneous calculation result may be output to the user device due to a resource failure or data falsification. Therefore, by providing the resource management apparatus 2 with the job calculation result monitoring unit 26, the reliability of the job calculation result is improved.

  FIG. 12 shows a functional block diagram of the resource management device 2 of the fourth embodiment. When the user apparatus 1 reserves the network, information for confirming the reliability of the calculation result is sent to the resource management apparatus 2, and the job calculation result monitoring unit 26 of the resource management apparatus 2 uses the information to execute the job execution result. The reliability of the calculation result can be improved by confirming as needed.

  FIG. 13 shows a flowchart for explaining the fourth embodiment. The scheduler unit 22 of the resource management device 2 receives a job reservation request from the user device 1 during the time period a to c (S30). The resource management device 2 issues a reservation certificate to the user device 1 (S31).

  The user device 1 makes a request for improving reliability in order to improve the reliability of the calculation result, and sends information for confirming the calculation result to the resource management device 2 (S32). The job is executed when the execution time is ac, and the calculation result is confirmed by the job calculation result monitoring unit 26 at any time (S33, S34, S35). If the calculation results are different, the calculation results are discarded (S36) and recalculation is performed. When the execution of the job exceeds the reservation time, the reservation time is extended or re-reserved in the scheduler unit 22.

  The first parameter input by the user device 1 is information for confirming the job execution result. For example, if it is known that the number of digits of the calculation result will be 4 digits or less, if the information becomes 4 digits or more by inputting the information, it can be discarded before the result is obtained. Other examples include even numbers, odd numbers, and a small number. As a result, the execution result can always be confirmed and the reliability of the execution result can be improved.

  Furthermore, by registering the computer resource with the wrong calculation result in the resource management device 2 and taking statistical data, whether the operator is an abnormality of the computer alone, an abnormality of the network, or the data falsified by someone, Alternatively, it is possible to know when it is concentrated, and it is possible to determine the reliability of the computer resources by preventing troubles or displaying them on the user device.

(Fifth embodiment)
As one method for improving the reliability of job results, it is possible to improve the reliability by reserving a plurality of resources redundantly. In this case, there are a method of providing a plurality of network resources with one computer resource, and a method of providing a plurality of computer resources and network resources.

  In the first case, when the resource management device 2 receives a reliability improvement request from the user device 1 when requesting resource reservation, the resource management device 2 refers to the job database 20 and provides a service class necessary for executing the job. A plurality of network resources are reserved in the scheduler unit 22.

  FIG. 14 is a diagram for explaining the first reliability improvement method. At this time, the network resource (path) to be reserved needs to satisfy the service class for both the active system (thick solid line) and the standby system (dashed line). For example, in the case of a service that is required to have a small delay variation, the resource management device 2 needs to provide a path whose delay does not change even when the route is switched.

  The resource management device 2 includes delay difference compensation means (not shown) after generating a plurality of paths, or a node device 6 having communication means with the delay difference compensation device (not shown). Send path delay compensation request. The node device 6 that has received the delay compensation request sends a delay measurement packet from both paths to the node device 5 that terminates both paths, and measures the delay difference between both paths. When the delay difference between the two paths is different, the same delay path can be provided by compensating the delay by the delay difference compensating means. One is an active path, and the other is a backup path. When a network failure occurs, the resource management device 2 sends a path switching request to the node device 6 on the network resource, thereby switching the route and improving the reliability of the network. be able to.

  The delay difference compensator may be a device that adjusts the transmission path of an optical signal dynamically or thermodynamically, or controls dispersion by providing dispersion through a variable dispersion device such as VIPA (Virtual Phased Array). Alternatively, the delay may be compensated by using a buffer, adjustment may be performed when a packet is read from the tag delay control flag, or the technology of Virtual Connection may be used.

  In the second case, when a reliability improvement request is received from the user device 1 when a resource reservation request is received, the resource management device 2 sends a message to the CRM (cluster resource management) or each cluster, Reserve multiple computer resources and reserve multiple paths connecting them.

  FIG. 15 illustrates a second method for improving reliability. The resource management device 2 periodically monitors the state of computer resources or CRM computer resources after the service is started. There are two methods for improving the reliability. One is a method for executing the same job in the active computer resource and the spare computer resource, and the other is a backup computer in which the active computer resource executes the job. The resource is a method of backing up the working computer resources at regular intervals.

  When there is no response from the computer resource 3-1 for a certain period, the resource management device 2 determines that a failure has occurred, and the resource management device 2 requests the node device 5 for switching the route to switch the computer resource 3-1. Is switched from the active computer resource 3-1 to the standby computer resource 3-2. When disconnection occurs, it is necessary to notify the opposite computer resource of the path switching. Further, the resource management device 2 notifies the user device 1 that the computer resource has been switched.

  Thereby, not only the reliability of network resources but also the reliability of computer resources can be improved.

  Further, when the job is a calculation, the resource management apparatus 2 takes the same job to execute the same job to collect the statistics of the job execution result, and outputs the higher result to the user apparatus as a more reliable value. You can also.

  In both cases, the resource management device 2 collects information about the computer resource with the wrong calculation result or information on the network resource in which the trouble has occurred from the user device 1, and notifies the operator of those resources of the error early on. Resource operators can avoid problems with computer resources and network resources.

  Also, by registering computer resources with the wrong calculation results in the resource management device 2 and collecting statistical data, the operator is concentrating on whether the computer is abnormal or a network problem, and when By analyzing the cause of the trouble or displaying the result on the user device, it is possible to know which resource has high reliability.

(Sixth embodiment)
The sixth embodiment is a program that, when installed on a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize functions corresponding to the resource management apparatus 2 of the first to fifth embodiments.

  By recording the program of this embodiment on a recording medium, the general-purpose information processing apparatus can install the program of this embodiment using this recording medium. Alternatively, the program of the present embodiment can be directly installed on the general-purpose information processing apparatus via a network from a server that holds the program of the present embodiment.

  Thereby, the function corresponding to the resource management apparatus 2 of this embodiment is realizable using a general purpose information processing apparatus.

  Note that the program according to the present embodiment includes not only a program that can be directly executed by a general-purpose information processing apparatus but also a program that can be executed by being installed on a hard disk or the like. Also included are those that are compressed or encrypted.

  According to the present invention, it is possible to improve the efficiency of the management function of network resources and computer resources, and therefore it is possible to provide convenience and improvement in service quality for both network administrators and users.

The whole block diagram of the conventional resource management system. The sequence diagram which shows the communication procedure between the conventional resource management apparatus and a user apparatus. The functional block diagram of the resource management apparatus in 1st embodiment. 1 is an overall configuration diagram of a resource management system for explaining a first embodiment. FIG. The sequence diagram for demonstrating 1st embodiment. The figure which shows an example of a job database. The figure which shows an example of a path | route database. The functional block diagram of the resource management apparatus of 2nd embodiment. The sequence diagram for demonstrating 2nd embodiment. The figure which shows the example of the job database used by 2nd embodiment. The figure for demonstrating 3rd embodiment. The functional block diagram of the resource management apparatus of 4th embodiment. The flowchart for demonstrating 4th embodiment. The figure explaining the 1st reliability improvement method. The figure explaining the 2nd reliability improvement method.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 User apparatus 2 Resource management apparatus 3, 3-1, 3-2 Computer resource 4 Network resource 5, 6 Node apparatus 20 Job database 21 User apparatus communication part 22 Scheduler part 23 Network control part 24 Path | route database 25 Execution time calculation part 26 Job calculation result monitoring part

Claims (9)

Resource management that receives a resource reservation request transmitted from a user apparatus, determines whether or not the corresponding resource can be allocated to the user apparatus, and if so, resource management that allocates a resource that satisfies the resource reservation request to the user apparatus In the device
A job database in which the correspondence between the identification information indicating the type of job and the service class information required for each job is recorded;
Resource information holding means for holding network resources and service class information indicating the quality of the network resources;
When a resource reservation request including job identification information is received from a user device, the job database is referred to, the service class information of the job is searched from the job identification information, and the network resource satisfying the requested quality of the service class is retrieved. A resource management apparatus comprising: means for determining based on service class information indicating the quality of network resources held in the resource information holding means and allocating to the user apparatus. The network resource includes a transmission line on the network or a path set in the transmission line,
The resource management apparatus according to claim 1, wherein the service class information includes information on a delay amount or a packet loss tolerance of the path or transmission path. The job database holds a correspondence relationship between a job process and a history of past execution times for each process of the job,
When a resource reservation request is received from a user device, the job database is referred to, and the execution time of the job is predicted by referring to the correspondence between the job and its execution time history based on the job identification information included in the resource reservation request. The resource management device according to claim 1, further comprising means for determining a resource satisfying the predicted execution time and allocating the resource to the user device. Means for comparing a service class of a network resource connected to a user device and a service class of a job being executed or scheduled to be executed by the user device;
If the network resource connected to the user device based on the comparison result by the comparing means cannot satisfy the required quality of the job being executed or scheduled to be executed by the user device, the network resource satisfying the required quality of the user device Means for searching for computer resources comprising:
When a computer resource having a network resource that satisfies the required quality of the user device is found based on a search result of the means for searching, a network resource serving as a connection path between the user device and the computer resource is allocated to the user device, The resource management apparatus according to claim 1, further comprising: means for transferring the job being executed or scheduled to be executed to the computer resource. Calculation result information collecting means for collecting calculation result information from the user device;
Means for determining a job execution result based on the calculation result collected by the calculation result information collecting means;
Means for prompting the user device to recalculate and discard the calculation result when an abnormality is found in the calculation result based on the determination result by the determination unit;
The resource management apparatus according to claim 1, further comprising means for extending or re-reserving the reservation time when the execution time of the job exceeds the reservation time in the calculation result information collecting process by the calculation result information collecting means. Means for allocating a plurality of redundant network resources satisfying the required quality for one job request upon receiving a reliability improvement request from the user device;
The resource management apparatus according to claim 1, further comprising means for switching to another network resource that bypasses the network resource in which the failure has occurred when a failure occurs in the network resource. Means for allocating a plurality of redundant computer resources in response to a request for one job upon receiving a reliability improvement request from a user device;
The resource management apparatus according to claim 1, further comprising means for switching a computer resource in which a failure has occurred to a computer resource in which no failure has occurred when a failure occurs in the computer resource. Resource management that receives a resource reservation request transmitted from a user apparatus, determines whether or not the corresponding resource can be allocated to the user apparatus, and if so, resource management that allocates a resource that satisfies the resource reservation request to the user apparatus In the resource management method executed by the device,
The resource management device records the correspondence between the identification information indicating the type of job and the service class information required for each job in the job database,
The resource information holding means holds network resources and service class information indicating the quality of the network resources,
When a resource reservation request including job identification information is received from a user device, the resource that satisfies the required quality of the service class retrieved by referring to the job database and searching for the service class information of the job from the job identification information A resource management method comprising: determining based on service class information indicating the quality of a resource held in the information holding means and allocating it to the user device.   A program that, when installed in a general-purpose information processing apparatus, causes the general-purpose information processing apparatus to realize a function corresponding to the function of the resource management apparatus according to any one of claims 1 to 7.

Images