JP4721057B2 - Data management system, data management method, and data management program - Google Patents

Description

  The present invention relates to a data management system, a data management method, and a data management program using a storage subsystem installed on a network, and in particular, between a plurality of storage subsystems in order to obtain business continuity when a storage subsystem fails. The present invention relates to a data management system, a data management method, and a data management program for backing up data in a storage subsystem to realize data multiplexing and to restore data when data is lost.

  In recent years, enterprise information systems that manage business operations of enterprises have shifted from computer systems that are collectively managed by large computers to distributed computer systems such as client server systems. Further, the enterprise information system is required to operate without a break, and a high operation rate, that is, high availability is required. For this reason, a storage subsystem used in a distributed computer system is usually equipped with a backup unit that stores emergency recovery data when a failure occurs in the primary unit.

  However, failures are not only due to problems of the unit itself, but may also occur due to natural disasters such as earthquakes or fires at the storage subsystem installation location, when the primary unit and backup unit are installed in the vicinity. Both units will be damaged by natural disasters. As a method for preparing for this, a method of performing remote copy of a volume by sending data stored in a volume of a storage subsystem installed at a certain site to another storage subsystem installed at a remote location, that is, A method of transferring write data from a host device between storage subsystems is adopted.

  There are three types of data transfer methods for executing remote copy between storage subsystems: synchronous mode, semi-synchronous order guarantee mode, and asynchronous mode.

  In the synchronous mode, information written to the master volume at the local site is synchronously sent to the replica volume at the remote site at a remote location, thereby ensuring the identity of both volumes. In this method, in response to an IO (IN, OUT) command from the host device, data is written to the master volume at the local site and data is written to the replication volume at the remote site, and the command completion report for both write processes is issued. When all of the above are completed, a completion report (command completion report) is returned to the host device. As a result, the command completion report from the remote site at a remote location will be confirmed each time, and there will be a delay due to the line distance and a delay due to the capacity of the line to which the data is to be transferred, degrading the response of the storage subsystem. Become.

  There is a semi-synchronous order guarantee mode as a method for improving the problem in the synchronous mode. This method is similar to the synchronous mode in that data is written to the master volume at the local site in response to an IO command from the host device, but the IO command from the host device is sent to the replication volume at the remote site. Before sending, the data is written in a semi-synchronous order guarantee buffer (hereinafter simply referred to as a buffer) in the local site. At this point, the process ends normally and an end report (command completion report) is returned to the host device.

  This buffer also transfers some of the IO instructions together in a single BOX (packaged) to the remote site, ensuring their order. At the remote site, a large amount of these packaged IO instructions arrives, but the order is guaranteed by processing in order based on the information written in the BOX.

  However, when a serious problem (disaster) occurs at the local site, the buffer capacity is lost up to the maximum, and the RPO (recovery point) becomes slightly worse. Further, as another problem, when the prepared buffer overflows, waiting for a buffer to occur is generated. Since the data in the buffer is governed by the delay of the line distance and the delay due to the capacity of the line to which the data is transferred, as in the synchronous mode, the movement to wait for the buffer to become free is the same as the synchronous type. End up.

  On the other hand, the asynchronous mode is a method of returning an end report (command completion report) to the host system when data is written to the master volume at the local site in response to an IO command from the host system. There is no extra response delay. In this method, difference information is managed by a difference data table that manages where the data is written on the master volume, and difference data based on the difference data table is transferred to the remote site. The difference data is sequentially sent at a speed corresponding to the capacity of the line.

  However, in the asynchronous mode, the order of the IO instructions is not written in the difference information shown on the data table, so if a serious problem (disaster) occurs in the local site in the presence of the difference information, the remote site Recovery using data from a certain replication volume cannot be guaranteed.

  In view of this, in the primary system as the local site, the amount of differential data generated each time data is written to the master volume is managed, and the secondary system as the remote site is managed according to the operating status of the system. Patent Document 1 discloses a data multiplexing system using an asynchronous mode that controls the difference data transfer frequency and records the last reception time of update data from the primary system in the secondary system. ing

Japanese Patent Laid-Open No. 10-198607

  However, in the system shown in Patent Document 1, when a problem (disaster) occurs, it is possible to grasp the data matching state between the primary system and the secondary system, and a serious problem (disaster) in the local site where there is difference information. If this happens, recovery using the data of the replication volume at the remote site cannot be guaranteed. In other words, in the synchronous mode and the semi-synchronous order guarantee mode, when a serious problem (disaster) occurs at the local site, data can be recovered to the state at the most recent time using a replication volume at the remote site. Asynchronous mode has the inconvenience that data can be recovered only to a state that goes back to the recorded time.

  However, when viewed from the viewpoint of the capacity of the line, when the data writing of the storage subsystem is viewed statistically, the peak value is 10 times or more of the average value. To transfer these write data using a communication line, in synchronous mode and semi-synchronous order guarantee mode, in order not to impair the instantaneous speed required for the storage subsystem, a line design equivalent to the disk performance Therefore, a large amount of line capacity is required as compared with the asynchronous mode. That is, the synchronous mode and the semi-synchronous order guarantee mode require an operation using a line having a capacity that is about one digit larger than that of the asynchronous mode.

  Therefore, the present invention provides a data management system and data management that can be operated using a low-capacity line and that can multiplex data among a plurality of storage subsystems using a conventional synchronous data transfer method. It is an object of the present invention to provide a method and a data management program.

In order to achieve the above object, a data management system of the present invention includes a main storage subsystem having a master volume for writing data related to a data write command from a host device, and a spare having a spare volume as a copy of the master volume. A data management system comprising a storage subsystem connected via a communication network,
The main storage subsystem has a function of executing data transfer processing for copying data to be written to the master volume to the spare volume so as to be interchangeable between a semi-synchronous order guarantee mode and an asynchronous mode, and the spare storage subsystem But,
A copy volume as a copy of the spare volume, a write mode control means for determining a transfer method of data sent from the main storage subsystem, either the semi-synchronous order guarantee mode or the asynchronous mode, and the write mode Synchronous write control means for writing the data determined to have been transferred in the semi-synchronous order guarantee mode by the control means to the duplicate volume in synchronization with the write to the spare volume, and the asynchronous by the write mode control means Update data writing means for asynchronously writing the data determined to have been transferred in the mode to the spare volume and writing to the replica volume, and the write mode control means receives the asynchronous mode from the main storage subsystem. The data transferred in is written A function of recording an address in the spare volume in a pre-installed update data table, and the update data writing means reads the update data from the spare volume based on the address recorded in the update data table, and It is characterized by writing in a replica volume .

  According to such a data management system, information transfer between the main storage subsystem and the spare storage subsystem when multiplexing data written by a data write command from the host device is performed in a semi-synchronous order guarantee mode and an asynchronous mode. And can be executed by switching.

If you like this, it is possible to back up a spare volume in the replication volume.

  In this way, the backup volume can be synchronized and backed up to the replication volume, and even if a problem occurs in the backup volume, it can be recovered using the replication volume.

In this way, it is possible to perform data replication asynchronously between the preliminary volume and copy volume, while a transfer between the master volume and the backup volume in asynchronous mode, if a failure occurs in the master volume Although recovery using a spare volume cannot be guaranteed, a duplicate volume separated from the spare volume can guarantee recovery.

  In this way, synchronous writing between the spare volume and the replica volume can be stopped when the data transfer method between the main storage subsystem and the spare storage subsystem has shifted to the asynchronous mode.

Next, the data management method of the present invention includes a main storage subsystem having a master volume for writing data related to a data write command from a host device, and a spare storage subsystem having a spare volume as a copy of the master volume. Are connected via a communication network, and the main storage subsystem executes data transfer processing for replicating data to be written to the master volume to the spare volume in a semi-synchronous order guarantee mode and an asynchronous mode so that they can be switched to each other In the data management system having a function, the spare storage subsystem that has received the data transferred from the main storage subsystem determines whether the data transfer method is the semi-synchronous order guarantee mode or the asynchronous mode. And, wherein when it is determined that the semi-synchronous order assurance mode, included can write to the transfer duplication volume data equipped in advance in synchronism with the writing to the pre volume, when it is determined that the asynchronous mode, the transfer The recorded data is written into the spare volume, the address in the spare volume to which the data is written is recorded in the update data table, and the update data is read from the spare volume based on the address written in the update data table. It writes to the said replication volume, It is characterized by the above-mentioned .

  According to such a data management method, the information transfer in the semi-synchronous order guarantee mode and the low-capacity line direction are performed between the main storage subsystem and the spare storage subsystem according to the increase / decrease in the data write command issue density of the host device. Multiplexing of data can be realized by switching and executing information transfer in the asynchronous mode.

If you like this, it is possible to back up the copy volume to synchronize the data to be written to the spare volume.

In this way, it is possible to back up the copy volume of data to be written to a spare volume asynchronously, while a transfer between the master volume and the backup volume in asynchronous mode, if a failure occurs in the master volume, Although recovery using a spare volume cannot be guaranteed, a duplicate volume separated from the spare volume can guarantee recovery.

Next, the data management program of the present invention includes a main storage subsystem having a master volume for writing data related to a data write command from a host device, and a spare storage subsystem having a spare volume as a copy of the master volume. Are connected via a communication network, and the main storage subsystem executes data transfer processing for replicating data to be written to the master volume to the spare volume in a semi-synchronous order guarantee mode and an asynchronous mode. In the spare storage subsystem in the data management system having the function to perform the processing, the process for determining the transfer method of the data transferred from the main storage subsystem, either the semi-synchronous order guarantee mode or the asynchronous mode Sense, the semi-synchronous synchronization writing process to write the copy volume of the write synchronization with write only as a replication of the preliminary volume of data transferred by the order assurance mode to the emergency volume, have been transferred in the asynchronous mode A process of recording the address in the spare volume to which the written data is written in the update data table, and an asynchronous write that reads the update data from the spare volume and writes it in the replica volume based on the address recorded in the update data table It is characterized by causing a computer to execute a loading process.

  According to such a data management program, similarly to the data management system and the data management method described above, the data written to the spare volume is synchronized and backed up to the duplicate volume, and the main storage subsystem and the spare storage subsystem are When the data transfer method shifts to the asynchronous mode, synchronous writing between the spare volume and the replica volume can be stopped.

Since the present invention is configured and functions as described above, even if a failure occurs in the master volume during the remote copy between the master volume and the spare volume in the asynchronous mode, the replication separated from the spare volume is possible. The volume can guarantee recovery.

  Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing the configuration of the main storage subsystem in the data management system of this embodiment. As shown in FIG. 1, the data management system according to this embodiment is configured such that a main storage subsystem 2 and a backup storage subsystem 3 are connected via a communication network.

  The main storage subsystem 2 includes a master volume 21 for writing data based on a writeIO (IN, OUT) command indicating data write from the host device 1, a buffer 22 for storing a writeIO command from the host device 1, A data table 23 indicating which part of the address of the master volume 21 has been updated by writing data accompanying the writeIO command from the host device 1, and data table updating means for updating the data table 23 as indicated by the writeIO command 24 and difference data reading means 25 for reading difference data from the master volume 21 based on the data table 23.

  Further, the main storage subsystem 2 further ensures that the write IO instruction stored in the buffer 22 is transferred to the spare storage subsystem 3 while ensuring this order, and the difference data read by the difference data reading means 25 is stored in the spare storage subsystem. A sending unit 26 having a function of transferring to the system 3 and a transfer mode control means 27 for monitoring the issue density of the writeIO command and controlling the function of the sending unit 26 according to the issue density, and the host device 1 Is provided with completion report means 28 for transmitting a command completion report indicating that data writing has been normally completed.

  As shown in FIG. 2, the buffer 22 stores a plurality of write IO instructions issued by the host device 1, packages the IO instructions in a guaranteed order, and the packaged IO instructions are stored in the spare storage subsystem. 3 has a function of outputting to the sending unit 26 for transfer. The data table update unit 24 has a function of updating the data table 23 in accordance with information indicated in the write IO command from the higher-level device 1.

  Also, a buffer 22 that stores a plurality of writeIO instructions with this order guaranteed, a sending unit 26 that transfers the writeIO instructions stored in the buffer 22 to the spare storage subsystem with this order guaranteed, and data for the master volume Semi-synchronous mode execution unit for executing information transfer in the semi-synchronous order guarantee mode by means of completion report means 28 having a function of transmitting an instruction completion report to the host device upon completion of writing and storing of data writing instruction to the buffer 2a is configured.

  Still further, a data table 23 indicating which part of the address of the master volume 21 has been updated by writing data accompanying the writeIO command, and data table updating means for updating the data table 23 as indicated by the writeIO command. 24, the differential data reading means 25 for reading the differential data stored in the address shown in the data table 23 from the master volume 21, and the completion of writing the data to the master volume 21 are transmitted to the host device. The completion report means 28 having a function and the sending unit 26 for transferring the difference data to the backup storage subsystem 3 constitute an asynchronous mode execution unit 2b for executing information transfer in the asynchronous mode.

  The transfer mode control means 27 has a function of controlling the semi-synchronous mode execution unit 2a and the asynchronous mode execution unit 2b to operate by switching, and this function is based on the issue density of IO instructions by the host device 1 The asynchronous mode execution unit 2b is operated at a timing that increases beyond a certain value. Specifically, the storage of the writeIO instruction in the buffer 22 is stopped, and the data table 23 is updated as indicated by the writeIO instruction received while the storage of the writeIO instruction in the buffer 22 is stopped. The data table updating unit 24 is controlled, the difference data stored at the address shown in the updated data table 23 is read from the master volume 21, the difference data reading unit 25 is controlled, and the difference data is stored in the spare storage. The sending unit 26 is controlled so as to be transferred to the subsystem 3.

  Furthermore, the semi-synchronous mode execution unit 2a is operated at a timing when the issuance density of the IO commands by the higher-level device 1 falls below a certain value. Specifically, storage of the write IO instruction in the buffer 22 is started, and the sending unit 26 is controlled so as to transfer the write IO instruction stored in the buffer 22 to the spare storage subsystem 3 while guaranteeing this order. The completion report means 28 is controlled to transmit an instruction completion report to the host device upon completion of data writing to the volume and completion of data write instruction storage in the buffer.

  That is, the transfer mode control means 27 stores the writeIO instruction in the buffer 22 and operates the system in the semi-synchronous order guarantee mode in which the stored writeIO instruction is transferred to the spare storage subsystem 3 while ensuring the order. The data table 23 indicating which part of the address of the master volume 21 has been updated is updated in accordance with the data writing to the master volume, and the differential data read from the master volume 21 based on this data table 23 is stored in the spare storage subsystem. 3 has a function of managing whether to operate in an asynchronous mode.

  Also, the transfer mode control means 27 switches to the system operation in the semi-synchronous order guarantee mode at the timing when the IO command issue density by the host device 1 falls below a certain value. By transferring data, it has a function of allowing the asynchronous mode and the semi-synchronous order guarantee mode to coexist. Furthermore, when the difference data becomes “0”, the asynchronous mode is terminated and the operation is shifted to the operation only in the semi-synchronous order guarantee mode.

  Here, the fixed value of the IO instruction issue density, which is a reference for determining whether the transfer mode control means 27 is in the semi-synchronous order guarantee mode or the asynchronous mode, is a preset threshold value or an IO instruction stored in the buffer 22. A value that exceeds the storage capacity of the buffer 22 can be considered.

  Completion reporting means 28 completes the instruction upon completion of data writing to the master volume 21 and completion of storing the IO instruction in the buffer 22 when the transfer mode control means 27 is operating in the semi-synchronous order guarantee mode. It has a function of transmitting a report to the host device 1 and a function of transmitting an instruction completion report upon completion of data writing to the master volume 21 when system operation is performed in the asynchronous mode.

  FIG. 3 is a block diagram showing the configuration of the spare storage subsystem in the data management system of the embodiment shown in FIG.

  As shown in FIG. 3, the backup storage subsystem 3 includes a backup volume 31 as a copy of the master volume 21 in the main storage subsystem 2 and a copy volume 32 that is a copy of the backup volume 31, and the main storage subsystem. 2 is provided with synchronous write control means 33 for synchronously writing the data written in the spare volume 31 to the duplicate volume 32 based on the writeIO command transferred from the second.

  Further, the spare storage subsystem 3 stops the synchronous write control means 33 when receiving the differential data from the main storage subsystem 2, writes the differential data only to the spare volume 31, and synchronously writes when the write IO command is received. Write mode control means 34 for operating the control means 33, an update data table 35 indicating which part of the spare volume 31 has been updated by differential data writing, and a copy volume 32 based on the update data table 35 Update data writing means 36 for reading unreflected update data from the backup volume 31 and writing it in the duplicate volume 32 is provided.

  The spare volume 31 is a volume for sequentially writing data in accordance with a packaged IO command transferred in the semi-synchronous order guarantee mode, or a volume for writing differential data transferred in the asynchronous mode.

  The write mode control means 34 monitors the transfer method between the main storage subsystem 2 and the spare storage subsystem 3, and the spare volume 31 and the duplicate volume 32 are triggered when this transfer method shifts to the asynchronous mode. A function of stopping synchronous writing performed between the update data table 35, a function of reflecting a writeIO instruction transferred thereafter in the update data table 35, and operating the update data writing means 36 based on the update data table 35. It has a function. In addition, it has a function of starting synchronous writing between the spare volume 31 and the replica volume 32 in response to the transition to the semi-synchronous order guarantee mode.

  That is, when the transfer method between the main storage subsystem 2 and the spare storage subsystem 3 shifts to the asynchronous mode, the write mode control means 34 executes the copy between the spare volume 31 and the duplicate volume 32 asynchronously, and performs semi-synchronization. When the mode is shifted to the order guarantee mode, it has a function of controlling to execute synchronous copying.

  The write mode control means 34 also transfers the update data specified before the switching to the replication volume 32 even after the transfer method between the main storage subsystem 2 and the spare storage subsystem 3 is switched to the semi-synchronous order guarantee mode. By writing, it has a function of coexisting asynchronous and synchronous copies. Furthermore, when the update data writing means 36 writes the update data to the replication volume 32 and the update data becomes “0”, the asynchronous data copy is terminated and a function of shifting to a synchronous only copy is provided.

  Next, the overall operation in the data management system of this embodiment will be described. Here, the data management method according to the present invention will also be described.

  FIG. 4 is a diagram showing an outline of operations in the data management system of the embodiment shown in FIGS. 1 and 3.

  Information transfer in the semi-synchronous order guarantee mode is executed between the main storage subsystem 2 and the spare storage subsystem 3, and the master volume 21 and the spare volume 31 are slightly shifted, but the main storage subsystem 2 Even when a serious problem (disaster) occurs, recoverable data is stored in the spare volume 31 of the spare storage subsystem 3. At this time, synchronous writing is operated between the spare volume 31 and the replica volume 32, and similarly recoverable data is stored in the replica volume 32.

  In this state, when the issue density of writeIO instructions from the host device increases and exceeds a certain value, the transfer method between the main storage subsystem 2 and the spare storage subsystem 3 is switched to the asynchronous mode. In response to this switching, synchronous writing between the spare volume 31 and the replica volume 32 of the spare storage subsystem 3 is stopped. Thereafter, since the transfer is performed in the asynchronous mode, the spare volume 31 cannot recover the master volume 21, but the replication volume 32 separated immediately before is secured as a recoverable volume.

  Next, when the issue density of the write IO command from the host device decreases and falls below a certain value, the transfer method between the main storage subsystem 2 and the spare storage subsystem 3 is switched to the semi-synchronous order guarantee mode. At this time, the asynchronous mode and the semi-synchronous order guarantee mode coexist.

  After that, the difference data does not increase, and all the difference data is transferred after a certain time, whereby the transfer in the asynchronous mode is completed.

  At this point, the master volume 21 and the spare volume 31 are slightly different as usual, but can be restored to the spare volume 31 even if a serious problem (disaster) occurs in the main storage subsystem 2. Data is stored.

  Next, in order to return from the asynchronous state to the synchronous state between the spare volume 31 and the replica volume 32, first, the data is written synchronously according to the writeIO command from the main storage subsystem 2, and the difference transferred during the asynchronous state Data is written asynchronously according to the update data table 35.

  While there is a difference between the spare volume 31 and the duplicate volume 32, the duplicate volume 32 is a volume that cannot recover the master volume 21, but there is no problem because the spare volume 31 becomes a recoverable volume.

  FIG. 5 is a sequence diagram showing an overall operation in the data management system of the embodiment shown in FIGS. 1 and 3.

  First, a write IO command is transmitted from the host device 1 to the main storage subsystem 2 (FIG. 5: step s100).

  Subsequently, in the main storage subsystem 2, the data accompanying this command is stored in the master volume 21 based on the write IO command from the host device 1 (FIG. 5: step s101, data writing step). Then, it is determined by the transfer mode control means 27 whether or not the write density of writeIO instructions has exceeded a certain value (FIG. 5: step s102, determination step). When it is determined that the issue density of the writeIO instruction does not exceed a certain value, the writeIO instruction is stored in the buffer 22, and the completion report means 28 completes the writing of data to the master volume 21 and the completion of storing the IO instruction in the buffer 22. The command completion report is transmitted to the upper level device 1 and the writeIO command stored in the buffer 22 is transferred to the spare storage subsystem 3 in a state in which this order is guaranteed (FIG. 5: Step s103, Step s104, Step s105, semi-synchronous transfer step).

  Thereafter, when it is determined that the issue density of the IO instructions transmitted from the host device 1 exceeds a certain value, the storage of the write IO instructions in the buffer 22 is stopped, and the write IO stored in the buffer 22 at this time is stopped. The instruction is transferred to the spare storage subsystem 3.

  Here, the contents of the above-described data writing process may be programmed and executed by the computer as a data writing process. Further, the content of the above-described semi-synchronous transfer process may be programmed and executed by the computer as a semi-synchronous transfer process.

  On the other hand, if it is determined that the IO command issue density by the host device 1 has exceeded a certain value, the completion report means 28 sends a command completion report upon completion of data writing to the master volume 21, and The data table 23 indicating which part of the address has been updated is updated according to the write IO instruction while the write IO instruction is being stored in the buffer 22, and the difference data is transferred from the master volume 21 based on the data table 23. The difference data is read out and transferred to the spare storage subsystem 3 (FIG. 5: step s106, step s107, step s108, step s109, asynchronous transfer step).

  Thereafter, when the write density of writeIO instructions from the host device falls below a certain value, and the difference data shown in the data table 23 of the main storage subsystem 2 falls below the certain value, reading of the difference data is stopped. Thus, the differential data specified at this point is transferred to the spare storage subsystem 3. Thereafter, the difference data does not increase, and all the difference data is transferred after a certain time.

  Here, the contents of the above-described asynchronous transfer process may be programmed and executed by the computer as an asynchronous transfer process. Further, in order to realize the operation as described above, the computer is configured to execute a switching execution process for switching between the semi-synchronous transfer process and the asynchronous transfer process according to the IO command issue density by the host device 1 and executing it. May be.

  In the spare storage subsystem 3, when a writeIO command is transferred from the main storage subsystem 2, data is written to the spare volume 31 based on this writeIO command, and in synchronization with this, the write volume 32 is also written to the duplicate volume 32. Data based on the command is written (FIG. 5: step s111, data replication step).

  Here, the contents of the data replication process described above may be programmed and executed by the computer as a synchronous writing process.

  When the difference data is transferred from the main storage subsystem 2, the difference data is written in the spare volume 31 (FIG. 5: step s112, difference data duplication process), and which part of the address of the spare volume 31 is updated. The update data table 35 shown is updated in accordance with the update of the spare volume 31 (FIG. 5: step s113), and the data read from the spare volume 31 based on this update data table is written to the replication volume 32 (FIG. 5). : Step s114, asynchronous writing process).

  Here, the contents of the above-described differential data copying process and asynchronous writing process may be programmed to be executed as a differential data writing process and an asynchronous writing process by a computer.

  As described above, according to the present embodiment, the main storage subsystem 2 temporarily stores the write IO command (data write command) from the host device 1 in the buffer 22 and collects several write IO commands together. The semi-synchronous order guarantee mode for transferring and the asynchronous mode for managing the difference information in the data table 23 and transferring the difference data based on the data table 23 can be used according to the line status. As a result, when data is remotely copied between the main storage subsystem 2 and the spare storage subsystem 3, the semi-synchronous order guarantee mode can be adopted even in a low-capacity line.

  Further, the spare storage subsystem 3 can execute a method of copying the spare volume 31 to the replica volume 32 by selecting synchronous or asynchronous according to the semi-synchronous order guarantee mode and the asynchronous mode.

It is a block diagram which shows the structure of the main storage subsystem in the data management system of one Embodiment in this invention. It is a figure which shows the structure of the buffer of the main storage subsystem in embodiment shown in FIG. FIG. 2 is a block diagram showing a configuration of a spare storage subsystem in the data management system of the embodiment shown in FIG. 1. It is a figure which shows the outline of operation | movement in the data management system of embodiment shown in FIG. It is a sequence diagram which shows the whole operation | movement in the data management system of embodiment shown in FIG.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Host apparatus 2 Main storage subsystem 3 Spare storage subsystem 2a Semi synchronous mode execution part 2b Asynchronous mode execution part 21 Master volume 22 Buffer 23 Data table 24 Data table update means 25 Differential data reading means 26 Sending part 27 Transfer mode control means 28 Completion report means 31 Spare volume 32 Replica volume 33 Synchronous write control means 34 Write mode control means 35 Update data table 36 Update data write means

Claims (3)

Data obtained by connecting, via a communication network, a main storage subsystem having a master volume for writing data related to a data write command from a host device and a spare storage subsystem having a spare volume as a copy of the master volume A management system,
The main storage subsystem has a function of executing data transfer processing for replicating data to be written to the master volume to the spare volume so as to be switchable between the semi-synchronous order guarantee mode and the asynchronous mode,
The spare storage subsystem is
A replica volume as a replica of the spare volume;
Write mode control means for determining a transfer method of data sent from the main storage subsystem, either the semi-synchronous order guarantee mode or the asynchronous mode;
Synchronous write control means for writing data determined to have been transferred in the semi-synchronous order guarantee mode by the write mode control means to the replicated volume in synchronization with writing to the spare volume ;
Update data writing means for writing data determined to have been transferred in the asynchronous mode by the write mode control means to the replica volume asynchronously with writing to the spare volume ;
The write mode control means has a function of recording an address in a spare volume to which data transferred in an asynchronous mode from the main storage subsystem is written in an update data table provided in advance,
The data management system, wherein the update data writing means reads update data from the spare volume based on an address recorded in the update data table and writes it to the duplicate volume . A main storage subsystem having a master volume for writing data related to a data write command from a host device and a spare storage subsystem having a spare volume as a copy of the master volume are connected via a communication network, and the main storage subsystem The storage subsystem is a data management system having a function of executing data transfer processing for copying data to be written to the master volume to the spare volume so as to be interchangeable between the semi-synchronous order guarantee mode and the asynchronous mode,
The spare storage subsystem that has received the data transferred from the main storage subsystem determines whether the data transfer method is the semi-synchronous order guarantee mode or the asynchronous mode;
Wherein when it is determined that the semi-synchronous order assurance mode, included can write to the transfer duplication volume data equipped in advance in synchronism with the writing to the pre volume,
If the asynchronous mode is determined , the transferred data is written to the spare volume, the address in the spare volume to which the data is written is recorded in the update data table,
A data management method, comprising: reading update data from the spare volume based on an address written in the update data table and writing the update data to the duplicate volume . A main storage subsystem having a master volume for writing data related to a data write command from a host device and a spare storage subsystem having a spare volume as a copy of the master volume are connected via a communication network, and the main storage subsystem The spare storage in the data management system having a function in which a storage subsystem executes data transfer processing for copying data to be written to the master volume to the spare volume in a semi-synchronous order guarantee mode and an asynchronous mode. In the subsystem,
A process of determining a transfer method of data transferred from the main storage subsystem, either the semi-synchronous order guarantee mode or the asynchronous mode;
The semi synchronous order guarantee mode data transferred in synchronization with the write to the preliminary volume writes the replication volume as a replication of the preliminary volume synchronization writing process,
A process for recording an address in the spare volume to which data transferred in the asynchronous mode is written in an update data table;
And asynchronous write processing for reading the update data from the spare volume based on the address recorded in the update data table and writing it to the replica volume ,
A data management program for causing a computer to execute.

Images