JP2009157880A - Server device and file system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a method, a device and a system storing all files and control information on a memory and accessing shared files at high speed by two or more computers. <P>SOLUTION: A virtual file system using a memory 1007 using no hard disk is constructed on server devices 101, 102 to share files. The file 1010 or the like on the memory 1007 of each server device is synchronized so that each server refers to the file as the same content. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to inter-server file sharing that enables a plurality of servers (computers) to share the same file using the memory of two or more servers.

  In a system that shares files, there is a means of placing a disk on each server and writing to each server.

Japanese Laid-Open Patent Publication No. 2003-84917 (Patent Document 1) describes a system that performs remote dual writing between storage control devices and enables exclusive control in units of files when sharing files from each host device. ing. That is, while the file management information is shared via the first network, the data transfer load is reduced by duplicating data between the remote dual writing devices of the two file servers via the second network (SAN). It does not prevent data transfer for writing.
JP 2003-84917 A

  As described above, in the invention described in Patent Document 1, exclusive control is performed on a remote disk and a remote computer by the first network, and data is double-written by the second network. You are sharing a file with the same content. However, since control information is exchanged in the first network and data is exchanged using the second network, there is a problem that the exclusion interval is long. Also, with this method, exclusive control can be performed only on a file basis.

  An object of the present invention is to provide a server device and a file system in which all files and control information are stored in a memory and a shared file is accessed at high speed by two or more computers.

  Another object of the present invention is to provide a system for recovering recorded information when it is lost when the server is down.

  Furthermore, it is another object of the present invention to provide a method that enables partial exclusion of files.

  The above and other objects and novel features of the present invention will be apparent from the description of this specification and the accompanying drawings.

  Of the inventions disclosed in the present application, the outline of typical ones will be briefly described as follows.

  A server device according to a representative embodiment of the present invention has a memory, an interface with a first network, and an interface with a second network, and the same data is duplicated outside the server device in the memory. The user program operating on the server device receives an update request for the file transmitted via the first network, and the user program operates on the server device to update the file. Requesting a daemon to perform update of the file in the server device, and requesting update of the replicated file outside the server device via the second network. .

  Another server device according to a representative embodiment of the present invention has a memory, an interface with a first network, and an interface with a second network, and the same data is stored in the memory outside the server device. A file to be copied is stored, a user program operating on the server device receives an update request to the file on the memory transmitted via the first network, and the user program updates the file to the server This daemon requests a daemon operating on the device, and this daemon checks the exclusive / access right information of the file in the server device and the exclusive / access right information of the copied file outside the server device via the second network. And the exclusive / access right information of the file in the server device and the exclusive / access right of the file outside the server device. When both the access right information update requirements are satisfied, the daemon updates the file on the memory in the server device, and the copied file outside the server device via the second network It is characterized by requesting the update.

  In this server device, a user program operating on the server device receives a file reference request transmitted via the first network, and the user program sends a file reference to a daemon operating on the server device. The daemon may request and refer to a file in the server device.

  Another server device according to a representative embodiment of the present invention has a memory, an interface with a first network, and an interface with a second network, and the same data is stored in the memory outside the server device. A file to be copied is stored, a user program operating on the server device receives an update request to the file on the memory transmitted via the first network, and the user program updates the file. Requesting the daemon operating above, the daemon checks the exclusive / access right information of the file in the server device and the exclusive / access right information of the copied file outside the server device via the second network, Exclusive / access right information of files in the server device, and exclusive / access of copied files outside the server device If any of the right information update conditions is satisfied, the daemon updates the file in the server device, and requests update of a replicated file outside the server device via the second network. It is characterized by that.

  This server device may be characterized in that file exclusion / access right information can be managed in units of files and in units of a predetermined capacity of files.

  In addition, in response to a confirmation request for exclusion / access right information of a predetermined file via the second network, the server device confirms the exclusion / access right information regarding the request, and the confirmation result is transmitted via the second network. It is good also as sending out.

  A file system according to an exemplary embodiment of the present invention includes a memory, an interface with a first network, a first server device having an interface with a second network, and an interface between the memory and the first network. And a second server device having an interface of the second network, and a file in which the same data is replicated in the memory of the second server is stored in the memory of the first server, via the first network The user program operating on the first server device receives the update request for the file transmitted in step 1, and the user program requests the daemon operating on the first server device to update the file. Confirm the exclusive / access right information of the file in the server device of 1 and, together with the second server The exclusive / access right information of the copied file of the device is confirmed with the daemon operating in the second server device via the second network, and the exclusive / access right information of the file in the first server device, and The daemon executes the update of the file in the first server device when both the exclusive / access right information update conditions of the copied file in the second server device are provided, and the second The update of the replicated file on the second server device is requested via the network.

  In this file system, the user program and daemon operate on the second server device, and when a failure occurs in the first server device, the second user program takes over processing via the first network. It may be characterized.

  In this file system, when a failure of the first server device is recovered, a daemon operating on the first server device is transferred to a daemon operating on the second server device as a file on the memory of the first server device. The synchronization state of the file on the memory of the second server device may be confirmed via the second network.

  Another file system according to the representative embodiment of the present invention includes a memory, an interface with a first network, a first server device having an interface with a second network, a memory, and a first network. And a second server device having an interface with the second network, a memory, an interface with the first network, and a third server device having an interface with the second network, A file in which the same data is replicated in the memory of the server device is stored in the memory of the first server device, and a file in which the same data is replicated in the memory of the third server device is stored in the second server device. An update request to the file stored in the memory and transmitted via the first network is sent to the first The user program that operates on the server device receives, the user program requests the daemon that operates on the first server device to update the file, and the daemon that operates on the first server device is in the first server device. Confirm the exclusive / access right information of the file, and confirm the exclusive / access right information of the duplicated file of the second server device with the daemon operating in the second server device via the second network. When the exclusive / access right information of the file in the first server device and the exclusive / access right information of the duplicated file in the second server device both have the update conditions, the first A daemon running on the first server device updates the file in the first server device, and replicates the file on the second server device via the second network. The user program operating on the second server device receives the update request for the file transmitted via the first network, and the user program updates the file on the second server device. The daemon that operates on the second server device confirms the exclusive / access right information of the file in the second server device, and, at the same time, the copied file of the third server device. The exclusive / access right information is confirmed with the daemon operating in the third server device via the second network, and the exclusive / access right information of the file in the second server device and the third server device If the exclusive / access right information of the copied file has both update conditions, a daemon that operates on the second server device is stored in the second server device. Run the update Airu and wherein the request to update the replicated files on the third server device via the second network.

  The effects obtained by typical ones of the inventions disclosed in the present application will be briefly described as follows.

  In the file system according to the representative embodiment of the present invention, it is possible to refer to each server as the same content by synchronizing the memory file of each server.

  Further, in the file system according to the representative embodiment of the present invention, since the file contents are synchronized with other servers, it is possible to perform high speed data recovery when taking over to another server due to a failure or the like. It becomes possible.

  Furthermore, in the file system according to the representative embodiment of the present invention, the file status can be maintained by resynchronizing the memory contents of the server that has not failed.

  Hereinafter, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

(First embodiment)
(Normal processing of the present embodiment)
FIG. 1 is a configuration diagram showing the configuration of the inter-server memory shared file system according to the first embodiment of the present invention. This inter-server memory shared file system includes a first server 101 and a second server 102, and host devices 121, 122, 123, and 124 that request data reference and update from these servers. Although only three servers (nth server in addition to the above two) are shown in the drawing, there may be four or more servers.

  The first server 101 and the second server 102 are connected by a first network (LAN) 1 and a second network 2. Further, it is connected to the host devices 121, 122, 123, 124 via the first network 1.

  The second network 2 may be a SAN (Storage Area Network) or the like, but is not necessarily limited to this. In the drawing, the network topology is expressed as a bus type, but it may be a ring type ring connection.

  In the present embodiment, each of the first server 101 and the second server 102 includes a user program 1002, a processor (CPU or DSP) (not shown) in which a daemon 1003 operates, and a memory 1007.

  The user program 1002 is activated by an access or activation instruction from a higher-level device or an interrupt by a timer (not shown). Via the daemon 1003, the user program 1002 executes exclusive processing, write processing, read processing, and change of access authority.

  In general, a “daemon” is a program that performs specific processing in the background in a multitasking environment. The daemon 1003 according to the present embodiment receives an instruction from the user program 1002, and accesses the memory 1007, transmits a command via the second network 2, receives a return value, and the like. Note that there is no distinction between primary and secondary between the daemon 1003 of the first server 101 and the daemon 1003 of the second server 102. The daemon 1003 of the first server 101 and that of the second server 102 can perform the same operation.

  The memory 1007 includes a work area 1008 which is an operation area of the processing apparatus, control data 1009 for storing information regarding exclusion and access authority, and a file 1010 to be referred to / updated.

  In order to access the memory 1007, the daemon 1003 includes a server reception processing module 1004, an update preparation / update confirmation module 1005, and a reference module 1006 as programs.

  The server reception processing module 1004 determines whether the command received from the user program 1002 is an update process or a reference process.

  If the command from the user program 1002 is an update process, an update preparation instruction is issued to the update preparation / update confirmation module 1005. At the same time, an update preparation instruction is issued to the update preparation / update confirmation module 1005 in the daemon 1003 of another server that is the object of file synchronization via the second network 2.

  When the server reception processing module 1004 receives the update processing and sends an update request to the update preparation / update confirmation module 1005, the update preparation / update confirmation module 1005 confirms the exclusive status and access right, and the file 1010 in the memory 1007 Is stored in the work area 1008 in the memory 1007. Similarly, when an update request is sent from another server sent via the second network 2, the update preparation / update confirmation module 1005 confirms the exclusive status and access right, and updates the file 1010 in the memory 1007. The data for storing is stored in the work area 1008 in the memory 1007.

  If the exclusive processing has already been performed on the desired file by another module, the update preparation / update confirmation module 1005 returns an error to the server reception processing module 1004. Further, when the update data can be stored in the work area 1008, the server reception process 1004 of the request source of the local server or another server is notified that there is no error and the writing to the work area is normally completed.

  After confirming that all the request destinations have written update data in the work area 1008, the server reception processing module 1004 instructs each update preparation / update confirmation module 1005 to confirm the update data.

  On the other hand, when one of the update request destinations returns that an error has occurred in writing to the work area, the server reception processing module 1004 instructs the update preparation / update confirmation module 1005 to cancel the update, and An error is returned to the requesting user program 1002.

  If the command from the user program 1002 is a reference process, the server reception processing module 1004 issues a reference instruction only to the reference module 1006 of its own server. The exclusive status and access right are confirmed with respect to the update preparation / update confirmation module 1005 of the own server, and then the contents of the file in the memory 1007 of the own server are referred to via the reference module 1006 to the requesting user program 1002. Returns a response.

  FIG. 2 is a flowchart showing processing until the user program 1002 of the first server 101 updates / references the file 1010 on the memory 1007 via the daemon 1003.

  In this figure, steps S203, S204, and S207 surrounded by a broken line indicate steps where the number of processing subjects increases as the number of servers in FIG. 1 increases.

  Hereinafter, this process will be described.

  The user program 1002 of the first server 101 is activated by a request related to the operation of the file 1010 from each host device or an interrupt process by a timer (step S200).

  Subsequently, a command and necessary parameters are issued from the user program 1002 of the first server 101 to the server reception processing module 1004 of the daemon 1003 (step S201). FIG. 3 is a conceptual diagram showing the structure of parameters (data) for controlling a daemon transmitted from the user program 1002 to the server reception processing module 1004.

  The necessary parameters here include a control flag 802 for controlling the daemon 1003, a file name 803, control data update contents 804, file write data 805, and the like.

  The control flag 802 is a parameter indicating control content. Here, a parameter of 1 is set for “write”, 2 for “read”, and 3 for “exclusive control”.

  A file name 803 is a file name of a file to be accessed.

  The control data update content 804 is a parameter indicating the content used when the control data 1009 is updated.

  The file write data 805 indicates the contents to be written to the file when the control flag 802 is “1”.

  With these parameters, an interface with the user program 1002 is established.

  Upon receiving this command, the server reception processing module 1004 interprets whether the command from the server reception processing module 1004 is related to file update processing, reference processing, or control data update, and processing The module that requests is determined (step S202). In the above example, if the control flag is 2, the server reception processing module 1004 determines that it is a processing request for the reference module 1006, and if it is other, it is for the update preparation / update confirmation module 1005.

  In the case of a file reference process (when the control flag 802 is 2), the server reception processing module 1004 uses the file name 803 to the reference module 1006 for the exclusive information of the file, the access authority, and the address in the memory 1007. The identification is requested (step S211).

  FIG. 4 shows the control data 1009 in the memory 1007 of each server used when confirming the access authority. The control data 1009 can be divided into control data 70 for the entire file and partial control data 71 for the file. Among them, the control data 70 for the entire file always exists, but the partial control data 71 for the file may not exist when partial exclusive control is not performed. Further, when partial control of the same file is performed at a plurality of locations, a plurality of partial control data 71 of the file can exist.

  First, the control data 70 for the entire file will be described.

  A file name 702 is a data field that represents a file name to be managed by the control data.

  The size 703 is a data field that represents the data amount of the target file managed by the control data.

  The file creator 704 is a data field that stores information indicating the creator of the target file managed by the control data.

  The access authority 705 is a data field that defines the access authority for the entire file managed by the control data.

  Creation, update, and reference date / time 706 is a data field indicating the date / time when the file was created.

  The memory head address 707 is a data field indicating from which address in the memory the head of the target file managed by the control data is stored.

  An exclusive flag 708 for the entire file is a data field that defines how to perform exclusive control of the file. In this data field, exclusive control of access from the file creator, exclusive control from other than the file creator, and exclusive control when two or more users access simultaneously are represented by bits. In this figure, 2 bits each, a total of 6 bits, but more parameters can be included.

  The exclusive flag 708 for the entire file is a flag that changes statically or dynamically according to the operation status of the file. Specifically, a setting that permits only referencing by a person other than the creator (static example), or a sixth bit (32 in decimal notation) is set to “1” when a problem such as atomicity occurs when updated at the same time. "(Set state)" and displaying the exclusive state (dynamic example) can be considered.

  If any one of the exclusive flags 708 of the entire file is “1” (set state), the corresponding process cannot be executed. For example, when the third bit (4 in decimal notation) is 1, a person other than the file creator cannot refer to the file. If any of the exclusive flags 708 is “0” (non-set state), it means that another user program is not accessing the file or access to the user program is permitted. The module 1006 or the update preparation / update confirmation module 1005 performs reference / write processing. At this time, the dynamically changing flag of the exclusive flag 708 of the entire file is set to “1”, and the dynamically changing flag at the end of the reference / write processing is set to “0” (release simultaneous access restriction). To do) should also be considered.

  The exclusive flag 708 and access authority 705 of the entire file are the core of the exclusive / access authority information.

  The partial exclusive control head address 709 is a data field that indicates the address of the stored memory when there is partial control data 71 of the file. When the partial control data 71 of the file does not exist, it may be expressed that the partial control data 71 does not exist by inputting NULL.

  Next, the partial control data 71 of the file will be described.

  The partial exclusion address 710 is a data field that indicates the start address of a memory area where partial exclusion is performed, and the partial exclusion size 711 is a data field that represents the amount of data to be subjected to partial exclusion.

  The partial exclusion flag 712 of the file is a data field that defines how to perform exclusive control on the target of partial exclusion. In this figure, the bit configuration of this field is associated with the exclusion flag 708 for the entire file, but the bit arrangement of the partial exclusion flag 712 for the file and the exclusion flag 708 for the entire file is not necessarily associated.

  When the partial control data 71 of the file exists, the reference module 1006 or the update preparation / update confirmation module 1005 also refers to the partial exclusion flag 712 of this file as one of the exclusion / access right information. . Note that the confirmation of this item may be performed in the same manner as the exclusive flag 708 for the entire file, and is omitted here.

  The partial exclusive control head address 713 is a data field that indicates the address of the stored memory when there is partial control data 71 of a different file. If there is no partial control data 71 of the file, NULL is input.

  Thus, by storing and managing partial control data 71 of different files in a daisy chain, it is possible to perform exclusive control finely.

  Returning to the description of FIG.

  In the file reference process, the reference module 1006 of each server mainly refers to the access authority 705, the entire file exclusion flag 708, and the file partial exclusion flag 712 from the control data 1009 in the memory 1007 of the server itself. To check for access authority.

  At this time, if access from the user program 1002 to the file is permitted from the exclusive information or the like (step S212: Yes), the server reception processing module 1004 completes I / O (input / output) completion. Return to 1002 (step S213). Conversely, when access to the file is not permitted (step S212: No), the server reception processing module 1004 returns an I / O error to the user program 1002 (step S206).

  In the case of file update processing (when the control flag 802 is 1), the server reception processing module 1004 outputs an update command and data corresponding to the write data 805 to the file in FIG. 3 to its own update preparation / update confirmation module 1005. At the same time, an update command and data are output to the update preparation / update confirmation module 1005 of the second server 102 via the second network 2. The update preparation / update confirmation module 1005 of each server has the access right of the request source (here, the server reception processing module 1004 of the first server 101) or the file to be accessed is placed under exclusive processing. Make sure that If access from the request source is not permitted, a message to that effect is returned to the server reception processing module 1004 of the first server 101 (step S203). In this case, the data is sent to the second server 102. However, when there are more servers, update commands and data may be output to all or a part of them.

  Thereafter, the update preparation / update confirmation module 1005 of each server executes a process of writing the received data to the work area 1008 of the memory 1007 (step S204). When the data cannot be written in the work area 1008 due to a lack of memory capacity or the like, the fact that an error has occurred is returned to the requesting server reception processing module 1004 that the data has been written without any problem.

  The server reception processing module 1004 of the first server 101 that is the request source verifies the return value of step S203 or step S204 received from each server (step S205). If there is a problem with even one return value (such as no access authority or an update file cannot be written to the work area 1008) (step S205: No), the server reception processing module 1004 depends on the content of the error. Returns an I / O error to the user program 1002 and ends the process (step S206).

  If all return values are appropriate (step S205: Yes), all servers have access authority. The server reception processing module 1004 updates the update preparation / update confirmation module 1005 of each server from the work area 1008 to the file 1010 (step S207). Specifically, it is conceivable to copy the data from the work area 1008 or rewrite the memory head address 707 in the control data 1009. The specific update method is a design matter.

  The success or failure of the update is returned as a return value from the update preparation / update confirmation module 1005 of each server to the server reception processing module 1004 of the first server that is the request source. With reference to these, the server reception processing module 1004 of the first server confirms whether the update of data of all the servers is appropriate (step S208). If the update has been properly performed on all servers (step S208: Yes), the server reception processing module 1004 of the first server notifies the user program 1002 of the completion of I / O (step S209). If an error has occurred in one server (step S208: No), the server reception processing module 1004 of the first server notifies the user program 1002 of an I / O error (step S210).

  When the control flag 802 in FIG. 3 is “3”, the control data 1009 such as partial exclusion in FIG. 4 is updated. The specific processing is the same as when the control flag 802 is “1”, but the processing target is not the file 1010 but the control data 1009. Also, the update data is not the write data 805 to the file, but the update contents 804 of the control data.

  In this way, it is possible to copy the same file on the memory of a different server without being conscious of the user program 1002. This advantage will be described below.

(Processing when a failure occurs in this embodiment)
FIG. 5 is a conceptual diagram showing the operation when a failure occurs according to this embodiment.

  In this figure, it is assumed that a failure occurs in the first server 101 and the processing is taken over by the second server 102. In addition, there is no distinction between primary and secondary between the first server 101 and the second server 102, and the description has been made with the configuration of two servers for convenience, but the same operation can be performed for three or more servers. It is possible.

  In the initial situation of this figure, the first server 101 and the user program 1002 of the first server 101 are operating normally. In addition, the second server 102 and the user program 1002 of the second server are running.

  In the memory 1007 in the first server 101, the file 2001 being edited and the file update history (journal) 2002 are recorded, and as described above, these are also recorded in synchronization with the memory 1007 of the second server 102. It shall be.

  When a failure occurs in the first server 101 such that the return value from the server reception processing module 1004 does not return or the return value always indicates an error, the user program 1002 of the first server 101 performs effective file processing. It becomes impossible to compensate.

  The user program 1002 of the first server 101 takes over the service to the user program 1002 waiting on the second server 102. At this time, the file 2001 being edited and the file update history (journal) 2002 are recorded in the memory 1007 of the second server 102 until just before a failure occurs in the first server 101. As a result, the user program 1002 on standby in the second server 102 only has to start the inherited process without performing the file inheritance, and the high-speed failure recovery is possible.

  Note that it is a design matter whether the user program 1002 of the second server 102 attempts to update the first server 101 that is the source of the failure after the takeover. In addition, what is assumed to be a failure (for example, whether the above shortage of memory capacity is regarded as a failure to be taken over) is left to the designer and is not considered here.

(Processing during failure recovery in this embodiment)
FIG. 6 is a conceptual diagram showing an operation in which the first server 101 recovers from a failure and copies a file from information in the memory 1007 of the second server 102.

  A memory recovery processing module 1011 is included in the daemon 1003 of each server for failure recovery. In the process of restarting the user program 1002 of the first server 101 where the failure has occurred, the daemon 1003 is started. However, at this stage, it is uncertain whether the contents of the memory, control data, files, etc. are appropriate.

  The memory recovery processing module 1011 of the first server 101 being restarted is stored in the memory 1007 of the second server 102 via the second network via the memory recovery processing module 1011 on the second server 102 that is another synchronization target. The control data 1009 and the file 1010 are replicated on the memory 1007 of itself. After replication is complete, the daemon restarts the service and reacts to responses from the higher-level user program 1002. Thereafter, if the problem is solved, the return value from the server reception processing module 1004 for the command of the user program 1002 becomes normal, and the failure recovery processing can be performed without particular awareness.

  In this way, if replication is performed on the daemon side, it is possible to perform failure recovery processing without taking action by the user program.

(Processing of the entire system at the time of reading in this embodiment)
FIG. 7 is a conceptual diagram illustrating file reading from a plurality of higher-level devices in the present embodiment to the server.

  As described in FIG. 2, the write is compensated for the synchronization of all servers. On the other hand, for reading, the reference module 1006 reads only from the memory 1007 of each server itself.

  Using this, a server to read from the host device is determined. In FIG. 7, the host device 121 reads the user program 1002 of the first server 101, the host device 122 and the host device 123 read the user program 1002 of the second server 102, and the host device 124 reads the user program 1002 of the third server 103. Is set to do.

  In this way, the contents written from any host device can be read from any host device. In addition, load concentration due to distribution of read requests is avoided.

  As described above, by requesting an update process for one server, data of all servers operating in cooperation can be updated synchronously. In addition, even when accessing the same synchronized file that exists on separate servers from a plurality of higher-level terminals, it is possible to prevent updates for access that interferes with synchronization, making it easy to maintain synchronization Can be achieved. Further, by providing partial control data 71 of the file in the control data, it becomes possible to perform partial access control of the file. In addition, when a failure occurs and is restored, the system can be restored without making the host user program aware of the failure by handling the failure with the daemon instead of the user program.

(Second Embodiment)
Next, a second embodiment of the present invention will be described.

  FIG. 8 is a conceptual diagram of a file duplication control method according to the second embodiment of the present invention.

  In the present embodiment, unlike the first embodiment, it is intended that not all servers have all memory files.

  FIG. 1 and FIG. 2 are used as the hardware and software configuration and control method in the present embodiment. However, efficient use of memory can be realized by distributing files stored in the server.

  An example of the control method of the present embodiment will be described with reference to FIG. In this example, it is assumed that the server is composed of three servers: a first server 501, a second server 502, and a third server 503.

  When the user program 1002 writes the file # 1 to the daemon 1003 of the first server 501, the server reception processing module 1004 of the daemon 1003 stores the file in the memory 1007 of itself and the memory 1007 of the second server 502. Update. On the other hand, the server reception processing module 1004 of the first server 501 does not copy, synchronize, or update files to the third server 503.

  When the user program 1002 writes the # 2 file to the daemon 1003 of the second server 502, the server reception processing module 1004 of the daemon 1003 stores the file in the memory 1007 of itself and the memory 1007 of the third server 503. Update. On the other hand, file replication / synchronization / update is not performed for the first server 501.

  When the user program 1002 writes the # 3 file to the daemon 1003 of the third server 503, the server reception processing module 1004 of the daemon 1003 stores the file in the memory 1007 of itself and the memory 1007 of the first server 501. Update. The second server 502 is not duplicated / synchronized / updated.

  By controlling as described above, the amount of backup files can be reduced, and the memory in each server can be used efficiently.

  Note that the above is merely an example of reducing the amount of file replication, and other control may be performed such that replication is performed on a server having a large free space in the memory.

  As mentioned above, the invention made by the present inventor has been specifically described based on the embodiments. However, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the scope of the invention. Needless to say.

  The present invention relates to a server that uses inexpensive and large-capacity memory in place of a disk device, and a system that includes a plurality of such servers and distributes processing among a plurality of servers (for example, use of content and shopping using a Wave Server) It can be used for high-speed file sharing under systems such as stock trading.

  Further, when the present invention is applied when the system is configured and the service is not interrupted, the recovery data is read from the memory by the standby server, so that high-speed system recovery is possible.

It is a block diagram which shows the structure of the memory sharing file system between servers which is the 1st Embodiment of this invention. It is a flowchart showing the process until the user program of this invention updates / references a memory via a daemon. It is a conceptual diagram which shows the structure of the parameter (data) for controlling the daemon transmitted with respect to a server reception processing module from a user program. It is a figure showing the control data in the memory of each server used when confirming access authority. It is a conceptual diagram showing the operation | movement at the time of the failure generation in the 1st Embodiment of this invention. It is a conceptual diagram showing the operation | movement which a 1st server recovers from a failure in the 1st Embodiment of this invention, and copies a file from the information of the memory of a 2nd server. It is a conceptual diagram explaining the file reading with respect to a server from the some high-order apparatus in the 1st Embodiment of this invention. It is a conceptual diagram about the file replication control method in the 2nd Embodiment of this invention.

Explanation of symbols

1 ... first network (LAN), 2 ... second network (SAN),
70 ... control data of the entire file, 71 ... partial control data of the file,
101 ... 1st server, 102 ... 2nd server, 103 ... 3rd server,
121, 122, 123, 124 ... host device,
501 ... first server, 502 ... second server, 503 ... third server,
1002 ... User program, 1003 ... Demon,
1004 ... Server reception processing module, 1005 ... Update preparation / update confirmation module,
1006 ... Reference module, 1007 ... Memory, 1008 ... Work area,
1009 ... control data, 1010 ... file,
1011 ... Memory recovery processing module,
2001 ... File being edited, 2002 ... File update history (journal),
3001 ... # 1 file, 3002 ... # 2 file, 3003 ... # 3 file.

Claims (10)

A server device having an interface with a memory, a first network, and an interface with a second network,
The memory stores a file in which the same data is copied outside the server device,
A user program operating on the server device receives an update request to the file transmitted via the first network;
The user program requests a daemon operating on the server device to update the file;
The server device, wherein the daemon updates the file in the server device, and requests the update of the replicated file outside the server device via the second network. A server device having an interface with a memory, a first network, and an interface with a second network,
The memory stores a file in which the same data is copied outside the server device,
A user program operating on the server device receives an update request to the file on the memory transmitted via the first network,
The user program requests a daemon operating on the server device to update the file;
The daemon confirms the exclusive / access right information of the file in the server device, and confirms the exclusive / access right information of the copied file outside the server device via the second network,
When the exclusive / access right information of the file in the server device and the exclusive / access right information of the file outside the server device have both update requirements, the daemon in the server device A server device that executes an update of the file on a memory and requests an update of the copied file outside the server device via the second network. In the server apparatus according to claim 1 or 2,
The user program operating on the server device receives a request to refer to a file on the memory transmitted via the first network,
The user program requests a daemon operating on the server device to reference the file;
The server device, wherein the daemon refers to the file in the server device. A server device having an interface with a memory, a first network, and an interface with a second network,
The memory stores a file in which the same data is copied outside the server device,
A user program operating on the server device receives an update request to the file on the memory transmitted via the first network,
The user program requests a daemon operating on the server device to update the file;
The daemon confirms the exclusive / access right information of the file in the server device and the exclusive / access right information of the copied file outside the server device via the second network,
When the exclusive / access right information of the file in the server device and the exclusive / access right information of the copied file outside the server device have both update conditions, the daemon A server device that executes update of the file in the device and requests update of the copied file outside the server device via the second network.   5. The server device according to claim 4, wherein the exclusive / access right information of the file enables management in file units and management of the file in a predetermined capacity unit. The server device according to claim 5,
In response to a request for confirmation of exclusive / access right information of a predetermined file via the second network, the daemon confirms the exclusive / access right information regarding the request, and sends a confirmation result via the second network. A server device. A first server device having an interface with a memory, a first network, and an interface with a second network, and a second server device having an interface with the memory, the first network, and an interface of the second network A file system including
A file in which the same data is copied to the memory of the second server device is stored in the memory of the first server device;
A user program operating on the first server device receives an update request to the file transmitted via the first network;
The user program requests a daemon operating on the first server device to update the file;
The daemon confirms the exclusive / access right information of the file in the first server device, and also sends the exclusive / access right information of the duplicated file of the second server device via the second network. Confirm with the daemon running in the second server device,
The exclusive / access right information of the file in the first server device and the exclusive / access right information of the duplicated file in the second server device both have update conditions. The daemon in the first server device updates the file in the first server device, and the replicated file on the second server device via the second network is updated. A file system characterized by requesting an update.   8. The file system according to claim 7, wherein when the user program and the daemon operate also on the second server device and a failure occurs in the first server device, the second user program executes the first user program. A file system characterized by taking over processing via a network.   9. The file system according to claim 8, wherein the daemon that operates on the first server device when the failure of the first server device is recovered is the daemon that operates on the second server device. A file system for confirming synchronization of a file on a memory of one server device and a file on a memory of the second server device via the second network. A first server device having an interface with a memory, a first network, an interface with a second network, an interface with a memory, an interface with the first network, and a second server device having an interface with a second network; A file system comprising a memory, an interface with a first network, and a third server device having an interface with a second network,
A file in which the same data is duplicated in the memory of the second server device is stored in the memory of the first server device, and the same data is duplicated in the memory of the third server device. Stored in the memory of the second server device,
A user program operating on the first server device receives an update request to the file transmitted via the first network;
The user program on the first server device requests a daemon operating on the first server device to update the file;
The daemon that operates on the first server device confirms the exclusive / access right information of the file in the first server device, and at the same time, exclusive / excludes the duplicated file of the second server device. Confirm the access right information with the daemon operating in the second server device via the second network,
The exclusive / access right information of the file in the first server device and the exclusive / access right information of the duplicated file in the second server device both have update conditions. The daemon operating on the first server device updates the file in the first server device, and the copied file on the second server device via the second network Request an update for
A user program operating on the second server device receives an update request to the file transmitted via the first network;
The user program on the second server device requests update of the file from a daemon operating on the second server device;
The daemon operating on the second server device confirms the exclusive / access right information of the file in the second server device, and at the same time, exclusive / excludes the copied file of the third server device. Confirm the access right information with the daemon operating in the third server device via the second network,
When the exclusive / access right information of the file in the second server device and the exclusive / access right information of the duplicated file in the third server device both have update conditions The daemon operating on the second server device updates the file in the second server device, and the replicated file on the third server device via the second network A file system characterized by requesting an update of a file.

Images