JP2011253400A - Distributed mirrored disk system, computer device, mirroring method and its program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a distributed mirrored disk system for restoring a disk device from a failure with small data transmission amounts and time.SOLUTION: A first computer device 10 writes data in a first data block 111, and a first mirroring control part 103 requests the writing of the data whose content is the same to a second computer device 20, and when receiving a reply indicating that the writing of the data has failed from a second mirroring control part 203 of the second computer device, writes difference information indicating the result via a difference information processing part 104 in a management information block 112 as that for the restoration of the second data block 211.

Description

  The present invention relates to a distributed mirrored disk system, a computer apparatus, a mirroring method, and a program therefor, and more particularly to a distributed mirrored disk system that can quickly recover a disk apparatus that constitutes a mirrored disk.

  In today's world, when computers or networks where computers are connected to each other are an important infrastructure as a foundation for companies and society, data loss is a socially important loss. Therefore, in servers and business terminals that operate business services, RAID (Redundant Arrays of Independent Disks) or RAIDs that operate a disk array that combines multiple disk devices as one virtual disk device. Often provides for data loss.

  Among a plurality of RAID standards, RAID 1 that can be most easily realized is also called mirroring. In this method, data having the same content is simply recorded on a plurality of disk devices. RAID 1 is not intended to increase the speed of data access, and only half of the total capacity of the disk device is substantially used, so the utilization efficiency and cost performance of the disk device are low. However, the fault tolerance of the stored data is increased.

  Here, in such a computer system that records data by RAID 1 (mirroring), for the purpose of disaster recovery (disaster recovery), it is possible to remotely connect via WAN (Wide Area Network) or the like. A system constituted by a plurality of computer devices installed on the ground and a disk device connected to each of the computer devices is called a “distributed mirrored disk system”. The distributed mirrored disk system synchronizes stored data between the respective disk devices by data communication.

  There are the following patent documents as technical documents related to this. Among them, Patent Document 1 discloses a distributed mirrored disk in which each computer constituting the system has a flag, and a computer that has recovered from a failure copies data corresponding to the flag until the flag matches another computer. The system is described.

  Patent Document 2 describes a replica volume resynchronization system in which both a master volume and a replica volume that are no longer in a synchronized state can be quickly resynchronized by providing a volume status identifier. In Patent Document 3, log information is recorded when one volume is in a suspended state, and based on this, the data is copied from the other volume after returning, thereby matching the recorded data among a plurality of volumes. It describes the storage system.

  Patent Document 4 describes an information system in which, when a failure occurs during the copying of data between storage apparatuses, it is not necessary to restart the application by switching the transmission destination of the write request for the data. Has been.

JP 2003-006015 A JP 2007-094656 A JP 2008-009707 A JP 2008-134988 A

  In RAID1, when one of the disk devices is recovered from a failure, it may be necessary to copy the entire storage contents of the disk device. Especially in a distributed mirrored disk system installed between remote locations via a WAN (Wide Area Network) etc., it is necessary to transmit a large amount of data via the network. Therefore, it takes a long time for recovery.

  The technique described in Patent Document 1 has a flag saving means for storing a flag for each of the volatile and nonvolatile storage means, and communicates flag on / off information between the computers every time data is written. It is realized by doing. However, since this communication is performed every time I / O for each block, which is a storage unit of disk data, is generated, the amount of data transmission required and the time required for it increase, resulting in trouble with network devices and disk devices. It is also a problem that there is a high possibility that this will occur. Further, when a network failure occurs here, there is a problem that a mismatch occurs between the volatile and non-volatile flag saving means, which may cause inconvenience.

  Further, in the technique described in Patent Document 3, in order to match the recording data among a plurality of volumes, log information is recorded when one volume is in a suspended state, and data is restored from the other volume after returning from the suspended state. Duplicate. In this case, the amount of data transmission during normal operation can be suppressed. However, there is a possibility that the “log information” mentioned here may be destroyed due to the occurrence of a failure or the like. In this case, it is necessary to copy the entire storage contents of the disk device. In this case, a technique capable of reducing the data transmission amount and time required for recovery of the disk device is not disclosed in any of the aforementioned Patent Documents 1 to 4.

  An object of the present invention is to provide a distributed mirrored disk system, a computer apparatus, a mirroring method, and a program thereof that can recover a disk apparatus from a failure with a small amount of data transmission and time.

  To achieve the above object, a distributed mirrored disk system according to the present invention is a distributed mirrored disk system comprising a first computer and a second computer device connected to each other via a network. The area in which the apparatus records differential information, which is information indicating that the first data block, which is an area for recording data, and data written in the data block has not been written in another computer apparatus. The first disk device in which the first management information block is secured, the first disk I / O unit that reads and writes data in the first data block, and the difference information is recorded in the first management information block A first differential information processing unit, and a first network I / O unit connected to a second computer device via a network; A first mirroring control unit that requests the second computer apparatus to write data having the same contents as the data written to the first data block, and is an area in which the second computer apparatus records data. A second disk device in which a second data block is secured, a second disk I / O unit for reading and writing data to and from the second data block, and a second computer connected to the first computer device via a network A network I / O unit, and a second mirroring control unit that writes data requested to be written from the first computer device to the second data block and returns the result of the writing to the first computer device. And the first management information block is an area accessible only from the first mirroring control unit. When the ringing control unit receives a reply from the second mirroring control unit indicating that the data requested to be written to the second computer device has failed to write, the difference information indicating the fact is used to restore the second data block. Is written in the first management information block.

  In order to achieve the above object, a computer apparatus according to the present invention is a computer apparatus that is connected to each other via a network and is connected to another computer apparatus to constitute a distributed mirrored disk system, for recording data. A disk device in which a data block that is an area for recording and a management information block that is an area for recording difference information that is information indicating that the data written to the data block has not been written to another computer apparatus; A disk I / O unit that reads / writes data from / to the data block, a differential information processing unit that records differential information in the management information block, a network I / O unit connected to another computer device via a network, and a data block The same data as the data written to the The management information block is an area that can be accessed only from the mirroring control unit, and the mirroring control unit writes the data requested to be written by another computer device to the data block. A function of returning the result of the writing to the requesting computer device, and when receiving a reply from the computer device indicating that the data requested to be written to the other computer device has failed to write, And a function of writing to the management information block for recovery of the second data block.

  In order to achieve the above object, a mirroring method according to the present invention is a distributed mirrored disk system configured such that a first computer device and a second computer device are connected to each other via a network. The first data block, which is an area for recording data on the first disk device provided in advance, and difference information which is information indicating that the data written in the data block has not been written in another computer device To secure the first management information block, which is an area accessible only from the first mirroring control unit for recording the data, and to record the data on the second disk device provided in advance in the second computer device A second data block that is an area of the first computer block is allocated to the first data block. The disk I / O unit writes data, and at the same time, the first mirroring control unit of the first computer device requests the second computer device to write data having the same contents as the data written to the first data block. The second mirroring control unit of the second computer apparatus writes the data that has received this write request to the second data block, and the second mirroring control unit of the second computer apparatus writes the result of the writing to the first data block. When the first mirroring control unit of the first computer apparatus receives a reply indicating that the writing has failed from the second computer apparatus, the difference information indicating that is sent to the second data block. The first management information block is written for recovery.

  In order to achieve the above object, a mirroring program according to the present invention is a distributed mirrored disk system configured by connecting first and second computer apparatuses to each other via a network, for recording data. Management information that is an area that is inaccessible from other programs for recording differential information that is information indicating that data blocks that are areas and data written to the data blocks are not written to other computer devices A procedure for writing data to the data block in the first computer device in which the block is reserved in advance, and a procedure for requesting the second computer device to write data having the same contents as the data written in the data block at the same time, this writing A procedure for receiving the result from the second computing device, and When receiving a reply indicating a failure written from the second computer device, characterized in that to execute a procedure for writing management information block difference information indicating that for the recovery of the second computer device.

  Since the present invention is configured to write difference information when data writing fails on the second computer device side as described above, not only the data transmission amount during normal operation is suppressed but also a failure occurs. Sometimes, it is possible to restore the disk device with a minimum amount of data transmission. Thus, it is possible to provide a distributed mirrored disk system, a computer apparatus, a mirroring method, and a program thereof that can recover the disk apparatus from a failure with a small amount of data transmission and time.

1 is an explanatory diagram showing a distributed mirrored disk system according to a first embodiment of the present invention. FIG. 2 is a flowchart showing an operation on the first computer device side when an application on the first computer device side issues an I / O request in the distributed mirrored disk system shown in FIG. 1. 4 is a flowchart showing an operation on the second computer device side when an application on the first computer device side issues an I / O request in the distributed mirrored disk system shown in FIG. 1. It is explanatory drawing which shows the content of the difference information between the data blocks when writing of the same data which the difference information processing part produced | generated by the difference information processing part shown in FIG. It is explanatory drawing which shows the content of the difference information between the data blocks at the time of the writing of the same data which the difference information processing part produced | generated by the difference information processing part shown in FIG. 3 is a flowchart showing an operation when a disk on the second computer device side is recovered from a failure that has occurred in the distributed mirrored disk system shown in FIG. 1. It is explanatory drawing shown about the reading of difference information and the copy of block data shown as step S403-404 of FIG. FIG. 7A shows a case where the same data as the block data 111 a is not written in the data block 211. FIG. 7B shows a state in which the difference information corresponding to this is acquired, and the block data on the data block at the address in the difference information is written as the block data at the same address on the data block. FIG. 7C shows a state in which the writing of block data is completed and the difference information corresponding to this is deleted. It is explanatory drawing shown about the structure of the dispersion | distribution mirrored disk system which concerns on the 2nd Embodiment of this invention. It is a continuation of FIG. 10 is a flowchart showing an operation on the first computer device side when an application of the first computer device issues an I / O request in the distributed mirrored disk system shown in FIGS.

(First embodiment)
Hereinafter, the structure of the 1st Embodiment of this invention is demonstrated based on attached FIG.
First, the basic content of the present embodiment will be described, and then more specific content will be described.
The distributed mirrored disk system 1 according to this embodiment is a distributed mirrored disk system configured by connecting first and second computer apparatuses to each other via a network. The first computer device 10 includes the first data block 111 that is an area for recording data and difference information that is information indicating that the data written in the data block has not been written in another computer device. The first disk device 12 in which the first management information block 112, which is a recording area, is secured, the first disk I / O unit 105 that reads and writes data in the first data block, and the first The first difference information processing unit 104 that records the difference information in the management information block, the first network I / O unit 106 connected to the second computer apparatus via the network, and the first data block A first mirroring control unit 103 that requests the second computer device to write data having the same contents as the data. The second computer device 20 includes a second disk device 22 in which a second data block 211 that is an area for recording data is secured, and a second disk I that reads / writes data from / to the second data block. / O unit 205, second network I / O unit 206 connected to the first computer device via the network, and data requested to be written by the first computer device are written to the second data block, And a second mirroring control unit 203 that returns a result of the writing to the first computer apparatus. When the first mirroring control unit 103 receives a reply from the second mirroring control unit indicating that the data requested to be written to the second computer device has failed to write, the second mirroring control unit displays difference information indicating that fact in the second information. Write to the first management information block for data block recovery. The first management information block is an area accessible only from the first mirroring control unit.

  Then, when the first mirroring control unit 103 restores the disk device 22 of the second computer device 20, the first mirroring control unit 103 checks whether or not the difference information exists in the management information block 112, and if the difference information exists. Based on this, the data that failed to be written is transmitted to the second computer device.

With this configuration, the distributed mirrored disk system of the present invention can recover the disk device from a failure with a small amount of data transmission and time.
Hereinafter, this will be described in more detail.

  FIG. 1 is an explanatory diagram showing a distributed mirrored disk system 1 according to the first embodiment of the present invention. The distributed mirrored disk system 1 is configured by connecting first and second computer apparatuses 10 and 20 to each other via a network 50. Incidentally, in FIG. 1, elements relating to the first computer apparatus 10 are indicated by “1”, and elements relating to the second computer apparatus 20 are indicated by “2".

  The first computer 10 includes a main arithmetic control unit 11 that is a main body that executes a computer program, a disk 12 that is a storage unit that records data, a communication unit 13 that communicates with other computers via a network 50, and a user. And input / output means 14 for presenting the processing result to the user. The disk 12 may be built in the casing of the first computer device 10 or may be externally attached.

  In the main calculation control means 11, the functions of the application 101, the I / O hook unit 102, the mirroring control unit 103, the difference information processing unit 104, the disk I / O unit 105, the network I / O unit 106, the node management unit 107, and the like. The section operates as a computer program. On the disk 12, areas such as a data block 111, which is an area for recording data, and a management information block 112 for recording information about the data recorded in the data block 111 are secured. The management information block 112 can read and write data only from the differential information processing unit 104. Details of each of these will be described later.

  The second computer device 20 has the same hardware and software configuration as the first computer device 10. The hardware configuration includes a main calculation control unit 21, a disk 22, a communication unit 23, and an input / output unit 24. The main calculation control means 21 includes an application 201, an I / O hook unit 202, a mirroring control unit 203, a difference information processing unit 204, a disk I / O unit 205, a network I / O unit 206, and a node management unit 207. The functional unit operates as a computer program. In the disk 22, areas such as a data block 211 and a management information block 212 are secured. Each of these has the same function as the element of the same name in the first computer apparatus 10.

  Thereafter, the first and second computer apparatuses 10 and 20 have the same hardware and software configurations, and therefore the first computer apparatus 10 side (application 101) is the trigger for the data write operation. A case will be described. That is, the principle of the operation of the distributed mirrored disk system 1 is that new data is first written on the first computer device 10 side and the same data is also written on the second computer device 20.

  The application 101 is software to be operated by the user on each computer apparatus. More specifically, it is software that performs functions such as a word processor, spreadsheet, database, web browser, e-mail, and the like.

  When the application 101 issues an I / O request to the disk 12 during operation, the I / O hook unit 102 captures this (referred to as “hooking an I / O request”), and this I / O It is determined whether the O request is a read request or a write request. If it is a read request, the designated data on the data block 111 is read via the mirroring control unit 103 and the disk I / O unit 105 and returned to the application 101. If it is a write request, the write data is passed to the mirroring control unit 103.

  The mirroring control unit 103 notifies the disk I / O unit 105 and the network I / O unit 106 of the write data passed by the I / O hook unit 102. When the network I / O unit 106 or the node management unit 107 notifies that the second computer device 20 has failed, the notification is transferred to the difference information processing unit 104. Further, when the application 101 tries to directly read / write data in the management information block 112, an I / O error is issued. That is, the management information block 112 can be accessed only for operations related to mirroring, and is not read or written by other operations.

  The difference information processing unit 104 generates difference information 112 a and stores the difference information 112 a in the management information block 112 when the same data cannot be written to the disk 11 of the first computer device 10 and the disk 22 of the second computer device 20. Write. When the disk 22 of the second computer device 20 is recovered from a failure, the corresponding write request information is output to the management information block 202 of the disk 12.

  The disk I / O unit 105 writes the write request data passed from the mirroring control unit 103 into the data block 111 of the disk 12. The success or failure of this write request is notified to the second computer device 20 via the network I / O unit 106.

  The network I / O unit 106 transmits the write request data passed from the mirroring control unit 103 to the network I / O unit 206 of the second computer device 20. In addition, the success or failure of the write request is received from the network I / O unit 206 of the second computer apparatus 20, and this is notified to the mirroring control unit 103. In the second computer device 20, the received write request data is notified to the mirroring control unit 203.

  The node management unit 107 performs node management of a computer configuration that performs disk mirroring, and notifies the mirroring control unit 103 when a failure occurs in the second computer device 20.

  FIG. 2 is a flowchart showing an operation on the first computer apparatus 10 side when the application 101 on the first computer apparatus 10 side issues an I / O request in the distributed mirrored disk system 1 shown in FIG. It is. In the flowchart of FIG. 2, the operation of the distributed mirrored disk system 1 is described with the first computer apparatus 10 side as an operation trigger. Therefore, in the operation described below, the application 201, the I / O hook unit 202, the differential information processing unit 204, and the management information block 212 on the second computer device 20 side are not used. This is why the elements corresponding to these are indicated by dotted lines in FIG.

  The I / O request issued by the application 101 is first hooked by the I / O hook unit 102 to determine whether this I / O request is a read request or a write request, and the determination result is mirrored. The unit 103 is notified (step S301).

  When the I / O request from the application 101 is a read request (step S301: NO), the I / O hook unit 102 notifies the mirroring control unit 103 of the read request and requests the data read (step S301). In step S320, the data read via the disk I / O unit 105 is returned to the requesting application 101 (step S321), and the process is terminated.

  When the I / O request from the application 101 is a write request (step S301: YES), the mirroring control unit 103 further operates the node management unit 207 of the second computer device 20 via the node management unit 107. It is inquired whether or not (step S302).

  If there is a response from the node management unit 207 that the second computer device 20 is operating (step S302: YES), the mirroring control unit 103 will send the first computer device 10 via the disk I / O unit 105. The data is written to the data block 111 (step S303), and the same data is passed to the mirroring control unit 203 of the second computer device 20 via the network I / O unit 106 to request writing to the data block 211. (Step S304), the result is waited (Step S305).

  Then, the mirroring control unit 103 determines whether or not the data writing to the data block 111 side is successful (step S306). If the data writing is successful (step S306: YES), the data writing to the data block 211 side is further performed. It is determined whether or not it is successful (step S307). If both have successfully written data, the result of data writing is returned to the requesting application 101 (step S308), and the process is terminated. This means that the same data is written in both the data blocks 111 and 211, and the difference information is not written in the management information blocks 112 and 212.

  In step S306, if the data writing to the data block 111 has failed (step S306: NO), the mirroring control unit 103 returns that fact to the requesting application 101 (step S308) and ends the process. This means that no new data is written in any of the data blocks 111 and 211, and there is no change in the management information blocks 112 and 212.

  If the data writing to the data block 111 was successful in step S306, but the data writing to the data block 211 side failed (step S307: NO), the difference information processing unit 104 is between the data blocks 111 and 211. (Step S312), this is written on the management information block 112 (step S313), and the mirroring control unit 103 returns that fact to the requesting application 101 (step S308) and ends the process.

  This means that new data has been written to the data block 111 but not written to the data block 211, so that a difference has occurred between the two, and it is necessary to write difference information. It is. Details of the difference information in steps S312 to S313 will be described later.

  In step S302 described above, if the node management unit 207 does not respond that the second computer device 20 is operating (step S302: NO), the mirroring control unit 103 passes through the disk I / O unit 105. Data is written in the data block 111 of the first computer apparatus 10 (step S310), and it is determined whether or not the data writing is successful (step S311).

  When the data writing to the data block 111 is successful in step S311 (step S311: YES), the difference information processing unit 104 generates difference information between the data blocks 111 and 211 (step S312). Writing to the management information block 112 (step S313), the mirroring control unit 103 returns a message to that effect to the requesting application 101 (step S308), and the process is terminated. This means that since new data has been written in the data block 111, a difference has occurred between the data blocks 111 and 211, so that it is necessary to write difference information.

  If the writing of data to the data block 111 has failed in step S311 (step S311: NO), the mirroring control unit 103 returns that fact to the requesting application 101 (step S308) and ends the process.

  FIG. 3 is a flowchart showing an operation on the second computer apparatus 20 side when the application 101 on the first computer apparatus 10 side issues an I / O request in the distributed mirrored disk system 1 shown in FIG. It is. When the network I / O unit 106 of the second computer device 20 receives the data write request in step S304 in FIG. 2 (step S351), the mirroring control unit 203 sends the data based on the request to the disk I / O unit 205. To the data block 211 (step S352), and returns the result to the first computer device 10 via the network I / O unit 206 (step S353). In the first computer device 10, the process following step S307 is performed according to the result.

  4 and 5 are explanatory diagrams showing the contents of the difference information between the data blocks 111 and 211 generated by the difference information processing unit 104 and written on the management information block 112 shown in steps S312 to S313 in FIG. It is. FIG. 4 shows a case where the same data has been successfully written to both of the data blocks 111 and 211, and FIG.

  Since the data blocks 111 and 211 are mirroring targets, in principle, the same data is always written to both. Therefore, as shown in FIG. 4, when block data 111a and 211a having the same contents are written in both data blocks 111 and 211, the difference information 112a is not written in the management information block 112.

  When the block data 111a is normally written only in the data block 111 and writing to the block data 211a having the same content as the data block 211 fails, as shown in FIG. 5, the data block 111 of the block data 111a is shown. The difference information 112 a including the above address and the fact that “data having the same content as this block could not be copied to the same block on the data block 211” is written in the management information block 112.

  FIG. 6 is a flowchart showing an operation when the disk 22 on the second computer device 20 side is recovered from the failure that has occurred in the distributed mirrored disk system 1 shown in FIG. Upon receiving the recovery command from the input / output unit 24, the mirroring control unit 203 first determines whether or not the recovery target disk 22 is a new disk (step S401), and if it is a new disk (step S401: YES). ) Requests the mirroring control unit 103 on the first computer device 10 side to transmit all the contents of the data block 111 on the first computer device 10 side, and copies all of the transmitted data to the data block 211. (Step S406) and the process is terminated.

  If the recovery target disk 22 is not a new disk (step S401: NO), the mirroring control unit 203 records the difference information 112a in the management information block 112 via the mirroring control unit 103 on the first computer device 10 side. It is determined whether or not it has been recorded (step S402), and if this is not recorded, the processing is terminated as it is (step S402: NO). The fact that the difference information 112a is not recorded means that there is no data to be newly copied because the contents of the data blocks 111 and 211 are the same.

  If the difference information 112a is recorded in step S402 (step S402: YES), the block data 111a recorded in the difference information is read and copied to the data block 211 to be restored (step S403). ). Then, the difference information 112a corresponding to the block for which copying has been completed is deleted (step S404).

  FIG. 7 is an explanatory diagram showing reading of the difference information 112a and copying of the block data 111a shown as steps S403 to 404 in FIG. FIG. 7A shows a case where the same data as the block data 111 a is not written in the data block 211. At the same time, the difference information 112 a corresponding to this is written in the management information block 112.

  In FIG. 7B, the corresponding difference information 112a is acquired, and the block data 111a on the data block 111 at the address in the difference information 112a is written as the block data 211a at the same address on the data block 211. Indicates the state. Step S403 is processing from the state of FIG. 7A to the state of FIG.

  FIG. 7C shows a state in which the writing of the block data 211a is completed and the corresponding difference information 112a is deleted. Step S404 is processing from the state of FIG. 7B to the state of FIG. Returning to FIG. 6, when the writing of one block data 211a is completed, the process proceeds to the next difference information 112a (step S405), and the processing from step S402 is repeated. If the process is completed for all the difference information 112a (step S402: NO), the recovery process is completed.

  The operation shown in FIG. 6 may be started when the mirroring control unit 103 on the first computer apparatus 10 side receives a recovery command from the input / output means 14. Since this operation is performed jointly by the mirroring control units 103 and 203, the trigger may be given from either one of the first and second computer apparatuses 10 and 20.

(Overall operation of the first embodiment)
Next, the overall operation of the above embodiment will be described. The mirroring method according to the present embodiment is a distributed mirrored disk system in which first and second computer devices are connected to each other via a network, and the first disk provided in advance in the first computer device. A first data block 111 that is an area for recording data on the apparatus and an area for recording difference information that is information indicating that the data written in the data block has not been written to another computer apparatus. A certain first management information block 112 is secured, and a second data block 211, which is an area for recording data on a second disk device provided in advance in the second computer device, is secured. Then, the first disk I / O unit of the first computer apparatus writes data to the first data block (FIG. 2: step S303), and at the same time, the same data as the data written to the first data block is written. The first mirroring control unit of the first computer device requests the second computer device to write the data of the contents (FIG. 2: step S304), and the received data is sent to the second computer device by the second computer device. Is written to the second data block (FIG. 3: step S352), and the result of this writing is returned by the second mirroring control unit of the second computer device to the first computer device (FIG. 3: FIG. 3). Step S353), the first mirroring control unit of the first computer device fails to write from the second computer device. When receiving a reply indicating that writes differential information indicating that the first management information block for the recovery of the second data blocks (Fig. 2: step S307,312~313).

Here, the above operation steps are programmed to be executable by a computer, and are executed by the first computer device 10 and the second computer device 20 which are computers that directly execute the respective steps. May be.
By this operation, this embodiment has the following effects.

  In the present embodiment, the difference information is recorded in the first management information block 112 only when data writing to the second data block 211 fails on the second computer device side. As a result, the amount of data transferred during normal operation can be reduced, and the occurrence rate of data mismatch between devices due to a network failure or disk failure can be reduced. Of course, even when the disk 22 of the second computer device 20 is restored, it is not necessary to copy the entire contents of the first data block 111, so that the data transfer amount can be reduced.

  Further, since the management information block 112 can be accessed only from the mirroring control unit 103 and cannot be accessed from other software (such as the application 101), the management information is intentionally destroyed, a disk failure, or a failure of the computer itself. It becomes possible to protect the difference information. Therefore, the entire contents of the first data block 111 are not copied, and the disk 22 of the second computer device 20 can be recovered with a small data transfer amount.

(Second Embodiment)
In addition to the first embodiment, a distributed mirrored disk system 401 according to the second embodiment of the present invention includes a third computer device 430 connected to the first and second computer devices via a network. It was set as the structure provided with. The third computer device 430 includes a third disk device in which a third data block 711 that is an area for recording data and a second management information block 712 that is an area for recording difference information are secured. 33, a third disk I / O unit 705 that reads / writes data to / from the third data block, a second difference information processing unit 704 that records difference information in the second management information block, and the network The third network I / O unit 706 connected to the first and second computer devices and the data requested to be written from the first computer device are written to the third data block, and the result of the writing is written to the first data block. A third mirroring control unit 703 that sends a reply to the computer apparatus. Then, the first mirroring control unit 503 requests the second and third computer apparatuses 420 and 430 to write data having the same contents as the data written in the first data block 511, and the data requested to be written is written. When a reply to the effect of failure is received from the second mirroring control unit 603, the difference information is written into the first management information block 512 and transmitted to the third computer apparatus 430, and the third mirroring control unit 703 The difference information received from the first computer device is written into the second management information block 712.

As a result, in this embodiment, the same effect as that of the first embodiment can be obtained. In addition, the difference information is destroyed and the entire contents of the data block must be copied to recover the disk device. There is also an effect that it is possible to reduce the possibility of being lost.
Hereinafter, this will be described in more detail.

  8 to 9 are explanatory diagrams showing the configuration of the distributed mirrored disk system 401 according to the second embodiment of the present invention. In FIGS. 8 to 9, the configuration is divided into two for convenience of space. The distributed mirrored disk system 401 is configured by connecting first, second, and third computer apparatuses 410, 420, and 430 to each other via a network 50. The hardware configuration of each of these computer devices is the same as that of the first and second computer devices 10 and 20 described as the first embodiment. Also, the configuration as software is roughly the same. Incidentally, in FIGS. 8 to 9, elements related to each of the first to third computer apparatuses 410 to 430 are indicated by “1 to 3”.

  Accordingly, the functional units constituting the first and second computer devices 410 and 420 according to the present embodiment are the same as the first and second computer devices 10 and 20 of the first embodiment in terms of hardware. Reference numbers may be used, and software and recorded data are reference numbers obtained by adding +400 to the reference numbers in the first embodiment.

  That is, in the main calculation control unit 11 of the first computer device 410, the application 501, the I / O hook unit 502, the mirroring control unit 503, the difference information processing unit 504, the disk I / O unit 505, and the network I / O unit 506 Each function unit such as the node management unit 507 operates as a computer program. The disk 12 has areas such as a data block 511 that is an area for recording data and a management information block 512 that records information about the data recorded in the data block 511.

  In the main calculation control means 21 of the second computer device 420, an application 601, an I / O hook unit 602, a mirroring control unit 603, a differential information processing unit 604, a disk I / O unit 605, a network I / O unit 606, a node Each functional unit such as the management unit 607 operates as a computer program. On the disk 22, areas such as a data block 611 and a management information block 612 are secured.

  The hardware configuration of the third computer device 430 includes a main arithmetic control unit 31, a disk 32, a communication unit 33, and an input / output unit 34. In the main arithmetic control means 31, each function such as an application 701, an I / O hook unit 702, a mirroring control unit 703, a differential information processing unit 704, a disk I / O unit 705, a network I / O unit 706, and a node management unit 707 is provided. The section operates as a computer program. On the disk 32, areas such as a data block 711 and a management information block 712 are secured.

  Each of these elements constituting the third computer apparatus 430 has the same function as the element of the same name in the first and second computer apparatuses 410 and 420.

  Hereinafter, only the difference in operation as software will be described in this specification. Further, the data write operation will be described in the case where the first computer apparatus 410 (application 501) is triggered. That is, the principle of operation of the distributed mirrored disk system 1 is that new data is first written on the first computer device 410 side, and the same data is also written on the second and third computer devices 420 and 430. To do.

  FIG. 10 shows an operation on the first computer device 410 side when the application 501 of the first computer device 410 issues an I / O request in the distributed mirrored disk system 401 shown in FIGS. It is a flowchart. The operations shown in this flowchart are substantially the same as the operations in the first embodiment shown in FIG. 2, and therefore the same operations are denoted by the same reference numerals.

  The only difference is the operation when the data written in the data block 511 on the first computer device 410 side is not written in the data blocks 611 and 711 of the second to third computer devices 420 to 430 ( (Corresponding to steps S312 to S313 in FIG. 2). The difference information generated here (step S801) is written in the management information block 512 of the first computer device 410 (step S802), and at the same time, it operates normally on the second and third computer devices 420 to 430. It is also written in the management information blocks 612 and 712 on the other side (step S803).

  That is, not only the contents of the data block but also the contents of the management information block are mirrored. For this reason, even if a problem occurs in the computer device holding the management information block and the difference information recorded there cannot be read, the difference information is read from the other device, and the case of the first embodiment It is possible to perform the same data recovery as (the operation described with reference to FIGS. 6 to 7). In other words, it is possible to reduce the possibility that the entire contents of the data block have to be copied.

  The present invention has been described with reference to the specific embodiments shown in the drawings. However, the present invention is not limited to the embodiments shown in the drawings, and any known hitherto provided that the effects of the present invention are achieved. Even if it is a structure, it is employable.

  About each embodiment mentioned above, it is as follows when the summary of the novel technical content is put together. In addition, although part or all of the said embodiment is summarized as follows as a novel technique, this invention is not necessarily limited to this.

(Appendix 1) A distributed mirrored disk system in which first and second computer devices are connected to each other via a network,
The first computer device comprises:
A first data block that is an area for recording data, and a first area that records differential information that is information indicating that the data written to the data block has not been written to another computer device. A first disk device in which a management information block is secured;
A first disk I / O unit for reading and writing data to and from the first data block;
A first difference information processing unit for recording the difference information in the first management information block;
A first network I / O unit connected to the second computer device via the network;
A first mirroring control unit that requests the second computer device to write data having the same content as the data written to the first data block;
The second computer device comprises:
A second disk device in which a second data block that is an area for recording data is secured;
A second disk I / O unit for reading and writing data to and from the second data block;
A second network I / O unit connected to the first computer device via the network;
A second mirroring control unit that writes data requested to be written from the first computer device to the second data block and returns a result of the writing to the first computer device;
The first management information block is an area accessible only from the first mirroring control unit;
When the first mirroring control unit receives a reply from the second mirroring control unit indicating that the data requested to be written to the second computer device has failed to write, the difference information indicating the fact is sent to the first mirroring unit. 2. A distributed mirrored disk system, wherein data is written to the first management information block for recovery of two data blocks.

(Supplementary Note 2) The first mirroring control unit
When the disk device of the second computer device is restored, it is checked whether or not the difference information exists in the management information block. If the difference information exists, writing has failed based on the difference information. The distributed mirrored disk system according to appendix 1, wherein the data is transmitted to the second computer device.

(Supplementary Note 3) A third computer device connected to the first computer device and the second computer device via the network is provided.
The third computer device comprises:
A third disk device in which a third data block which is an area for recording data and a second management information block which is an area for recording the difference information are secured;
A third disk I / O unit for reading and writing data to and from the third data block;
A second difference information processing unit for recording the difference information in the second management information block;
A third network I / O unit connected to the first and second computer devices via the network;
A third mirroring control unit that writes data requested to be written from the first computer device to the third data block and returns a result of the writing to the first computer device;
The first mirroring control unit requests the second and third computer devices to write data having the same contents as the data written to the first data block, and that the requested data has failed to be written. When the reply is received from the second mirroring control unit, the difference information is written to the first management information block and transmitted to the third computer device.
The supplementary note 1 is characterized in that the third mirroring control unit writes the difference information received from the first computer device into the second management information block for recovery of the second data block. The described distributed mirrored disk system.

(Supplementary Note 4) A computer apparatus that is connected to each other via a network and connected to another computer apparatus to form a distributed mirrored disk system,
A data block that is an area for recording data and a management information block that is an area for recording differential information that is information indicating that the data written to the data block has not been written to another computer device are secured. Disk unit,
A disk I / O unit for reading and writing data in the data block;
A difference information processing unit for recording the difference information in the management information block;
A network I / O unit connected to another computer device via the network;
A mirroring control unit that requests other computer devices to write data having the same content as the data written to the data block, and
The management information block is an area accessible only from the mirroring control unit,
The mirroring control unit
A function of writing data requested to be written from another computer device to the data block, and returning a result of the writing to the requesting computer device;
A function of writing, in the management information block, difference information indicating that the data has been rewritten to the second data block when a reply indicating that the data requested to be written to another computer apparatus has failed to be written is received from the computer apparatus. A computer apparatus comprising:

(Supplementary Note 5) In a distributed mirrored disk system configured by connecting first and second computer devices to each other via a network,
The first data block, which is an area for recording data on the first disk device provided in advance in the first computer device, and the data written in the data block were not written in another computer device. A first management information block that is an area accessible only from the first mirroring control unit for recording difference information that is information indicating
Securing a second data block which is an area for recording data on a second disk device provided in advance by the second computer device;
A first disk I / O unit of the first computer device writes data to the first data block;
At the same time, the first mirroring control unit of the first computer device requests the second computer device to write data having the same contents as the data written to the first data block,
The second mirroring control unit of the second computer device writes the data that has received this write request to the second data block,
The result of this writing is sent back to the first computer device by the second mirroring control unit of the second computer device,
When the first mirroring control unit of the first computer apparatus receives a reply indicating that the writing has failed from the second computer apparatus, the difference information indicating the fact is used for recovery of the second data block. And writing into the first management information block.

(Supplementary Note 6) In a distributed mirrored disk system in which first and second computer devices are connected to each other via a network,
Inaccessible from other programs for recording the data block, which is an area for recording data, and differential information, which is information indicating that the data written to the data block has not been written to another computer device In the first computer device in which a management information block that is a secure area is secured in advance,
Writing data into the data block;
At the same time, a procedure for requesting the second computer device to write data having the same contents as the data written to the data block,
A procedure for receiving the result of the writing from the second computer device;
And when a reply to the effect of writing failure is received from the second computer device, a procedure for writing the difference information indicating that in the management information block for recovery of the second computer device is executed. Mirroring program.

  The present invention can be applied to the use of protecting data by mirroring. In particular, it is suitable for applications such as data backup to a remote location via a network, or a data protection system for clustering that increases system availability.

DESCRIPTION OF SYMBOLS 1 Distributed mirror disk system 10, 20, 410, 420, 430 Computer apparatus 11, 21, 31 Main operation control means 12, 22, 32 Disk 13, 23, 33 Communication means 14, 24, 34 Input / output means 50 Network 101, 201, 501, 601, 701 Application 102, 202, 502, 602, 702 I / O hook unit 103, 203, 503, 603, 703 Mirroring control unit 104, 204, 504, 604, 704 Difference information processing unit 105, 205 , 505, 605, 705 Disk I / O unit 106, 206, 506, 606, 706 Network I / O unit 107, 207, 507, 607, 707 Node management unit 111, 211, 511, 611, 711 Data block 111a, 211a Block data 112, 212, 512, 612, 712 Management information block 112a Difference information

Claims (6)

A distributed mirrored disk system configured by connecting first and second computer apparatuses to each other via a network,
The first computer device comprises:
A first data block that is an area for recording data, and a first area that records differential information that is information indicating that the data written to the data block has not been written to another computer device. A first disk device in which a management information block is secured;
A first disk I / O unit for reading and writing data to and from the first data block;
A first difference information processing unit for recording the difference information in the first management information block;
A first network I / O unit connected to the second computer device via the network;
A first mirroring control unit that requests the second computer device to write data having the same content as the data written to the first data block;
The second computer device comprises:
A second disk device in which a second data block that is an area for recording data is secured;
A second disk I / O unit for reading and writing data to and from the second data block;
A second network I / O unit connected to the first computer device via the network;
A second mirroring control unit that writes data requested to be written from the first computer device to the second data block and returns a result of the writing to the first computer device;
The first management information block is an area accessible only from the first mirroring control unit;
When the first mirroring control unit receives a reply from the second mirroring control unit indicating that the data requested to be written to the second computer device has failed to write, the difference information indicating the fact is sent to the first mirroring unit. 2. A distributed mirrored disk system, wherein data is written to the first management information block for recovery of two data blocks. The first mirroring controller is
When the disk device of the second computer device is restored, it is checked whether or not the difference information exists in the management information block. If the difference information exists, writing has failed based on the difference information. 2. The distributed mirrored disk system according to claim 1, wherein the transmitted data is transmitted to the second computer device. A third computer device interconnected with the first and second computer devices via the network;
The third computer device comprises:
A third disk device in which a third data block which is an area for recording data and a second management information block which is an area for recording the difference information are secured;
A third disk I / O unit for reading and writing data to and from the third data block;
A second difference information processing unit for recording the difference information in the second management information block;
A third network I / O unit connected to the first and second computer devices via the network;
A third mirroring control unit that writes data requested to be written from the first computer device to the third data block and returns a result of the writing to the first computer device;
The first mirroring control unit requests the second and third computer devices to write data having the same contents as the data written to the first data block, and that the requested data has failed to be written. When the reply is received from the second mirroring control unit, the difference information is written to the first management information block and transmitted to the third computer device.
The third mirroring control unit writes the difference information received from the first computer device into the second management information block for recovery of the second data block. Distributed mirrored disk system described in 1. A computer device that is connected to each other via a network and connected to another computer device to form a distributed mirrored disk system,
A data block that is an area for recording data and a management information block that is an area for recording differential information that is information indicating that the data written to the data block has not been written to another computer device are secured. Disk unit,
A disk I / O unit for reading and writing data in the data block;
A difference information processing unit for recording the difference information in the management information block;
A network I / O unit connected to another computer device via the network;
A mirroring control unit that requests other computer devices to write data having the same content as the data written to the data block, and
The management information block is an area accessible only from the mirroring control unit,
The mirroring control unit
A function of writing data requested to be written from another computer device to the data block, and returning a result of the writing to the requesting computer device;
A function of writing, in the management information block, difference information indicating that the data has been rewritten to the second data block when a reply indicating that the data requested to be written to another computer apparatus has failed to be written is received from the computer apparatus. A computer apparatus comprising: In a distributed mirrored disk system configured by connecting first and second computer apparatuses to each other via a network,
The first data block, which is an area for recording data on the first disk device provided in advance in the first computer device, and the data written in the data block were not written in another computer device. A first management information block that is an area accessible only from the first mirroring control unit for recording difference information that is information indicating
Securing a second data block which is an area for recording data on a second disk device provided in advance by the second computer device;
A first disk I / O unit of the first computer device writes data to the first data block;
At the same time, the first mirroring control unit of the first computer device requests the second computer device to write data having the same contents as the data written to the first data block,
The second mirroring control unit of the second computer device writes the data that has received this write request to the second data block,
The result of this writing is sent back to the first computer device by the second mirroring control unit of the second computer device,
When the first mirroring control unit of the first computer apparatus receives a reply indicating that the writing has failed from the second computer apparatus, the difference information indicating the fact is used for recovery of the second data block. And writing into the first management information block. In a distributed mirrored disk system configured by connecting first and second computer apparatuses to each other via a network,
Inaccessible from other programs for recording the data block, which is an area for recording data, and differential information, which is information indicating that the data written to the data block has not been written to another computer device In the first computer device in which a management information block that is a secure area is secured in advance,
Writing data into the data block;
At the same time, a procedure for requesting the second computer device to write data having the same contents as the data written to the data block,
A procedure for receiving the result of the writing from the second computer device;
And when a reply to the effect of writing failure is received from the second computer device, a procedure for writing the difference information indicating that in the management information block for recovery of the second computer device is executed. Mirroring program.

Images