JPH10254787A - Guarantee processor for equivalence of duplex volume - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a means for reducing an area requiring a whole equivalence guaranteeing processing to shorten time by executing the equivalence guaranteeing processing on the area in which the equivalence guaranteeing processing can be executed by an application program with the logic of the application program with respect to the equivalence guaranteeing processor of a duplex volume. SOLUTION: A processor is provided with an equivalence guaranteeing processing means by individual applications 5 executing the equivalence guaranteeing processing in the area based on an area management table for equivalence guaranteeing processing object by individual applications 2 on the application program managing the completion recognition of writing and an equivalence guaranteeing processing selection means 3 selecting the equivalence guaranteeing processing means by individual applications 5 or a whole equivalence guaranteeing processing means 4 as the equivalence guaranteeing processing for respective volume areas based on a volume state management table 1 and the area management table of equivalence processing object by individual applications 2.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] 1. Field of the Invention [0002] The present invention relates to a duplicate volume equivalence assurance processing apparatus.

[0002]

2. Description of the Related Art The number and capacity of file devices for storing data have increased as computer systems have become more complex, larger, and continuously operated. It is required that the repair of the failure can be performed in parallel while continuing the operation so as not to affect the operation. One of the methods uses a duplex volume or file duplication technique.

[0003] That is, a file for writing data is prepared doubly in another volume, and the same data is simultaneously written to two different volumes. When reading these data, one of the data stored in duplicate is read. Since the same contents of data are stored in both volumes, that is, the equivalence of the data is guaranteed, so that it is sufficient that either volume can be read normally.

When any of the volumes becomes incapable of reading or writing due to a hardware failure, the failed device is disconnected, and with respect to the volume mounted therein, the duplication is stopped and only the volume of the normally operating device is stopped. By reading and writing, the system using the volume can continue operation while repairing the separated device in parallel. However, when the system is suddenly stopped due to a power failure or the like, the data in the duplicated volumes is not guaranteed to be equivalent.

Conventionally, the operation of a duplicate volume is performed as follows in a normal operation state and an abnormal state as described above. First, when stopping the system normally, for each of the duplicated volumes mounted on the system, if the duplicated volume is being updated, complete the update and ensure the equivalence of the duplicated volume,
The equivalence assurance completion status is displayed and set in the volume status management table 1 on the non-volatile medium, the system is stopped, and the power is turned off.

When the system is started, as shown in FIG. 8, the volume status management table 1 is checked for each of the duplicated volumes mounted on the system in steps S81 and S82, and the following processing is performed depending on the volume status. .

When the equivalency assurance completion status is set After the duplex operation status is set in the volume status management table, the duplex volume operation is started. That is, reading is performed from one of the volumes, and writing is performed to both volumes (step S85).

[0008] In the case where the duplex operation state is set, this is the state at the time of system startup after the system is stopped due to forced power cut or power failure during system operation. In this case (step S83), one of the volumes is set as the copy source volume, the other volume is set as the copy destination volume, and the equivalency assurance processing and the copy source volume name are set in the volume status management table 1, and then the copy source volume is set. Is copied to the copy destination volume (this is hereinafter referred to as a full equivalence assurance process). Reading and writing to the volume is possible even during the overall equivalence guarantee processing,
Since this is not a duplex state, if a failure occurs in the copy source volume, this volume cannot be used. When the entire equivalence assurance process is completed, the duplex operation status is set in the volume status management table 1 (step S85).

[0009] When the equivalency assurance processing in progress state is set This is the state at the time of system startup after the system is stopped before the complete equivalence assurance processing is completed (step S84).
The copy source volume and the copy destination volume are determined based on the volume state management table 1, and the entire equivalence assurance process is started. When the full equivalence assurance processing is completed, the duplex operation status is set in the volume status management table (step S8).
5).

[0010]

As described above, when the system is started after the system has been stopped due to a forced power cut or a power outage, it is necessary to perform full equivalence assurance processing for all the duplicated volumes in the system. Become. In the full equivalence assurance processing, since the entire area of the volume is copied, the I / O
The effect of the O load on business efficiency cannot be ignored. If it is attempted to perform the full equivalency assurance processing within the range that does not affect the business, it will take a long time to complete the full equivalency assurance processing of all the duplicated volumes in the system.
This is a problem from the viewpoint of reliability. This is a serious problem for systems with large amounts of data, such as large databases. For example, when 500 GB data is subjected to full equivalence assurance processing at a speed of 5 MB / sec, the time required for simple calculation is 28 hours, and the business is stopped for one day or more.

A method is conceivable in which an update log (which block of which volume is to be updated) is always collected before updating the duplicated volume and is made non-volatile. In this case, even if the system is stopped due to a forced power cut or a power failure, it is possible to know which block of which volume was being updated, so that equality of only that portion may be guaranteed. However, this method requires the collection of an update log in addition to the data update, so that the I / O response of all the update processes deteriorates about twice.

That is, as described above, there is a problem that the efficiency of the system operation and the business operation of the full equivalence assurance processing of the duplicated volume when the system is started after the abnormal termination of the system due to the sudden power failure such as a power failure has a problem. There was a need for a solution.

However, the atomicity of a series of procedures consisting of a plurality of data updates (atomicity, that is, whether all data are updated or not at all, is included in the application program) In some cases, the update log is collected by the logic of the application itself (for example, transaction processing by a database management system) in order to guarantee the above.
The update log collected by the application program contains the address of the data that is being updated, and if this is used, the part that contains the data that may be being updated even when the system stops abnormally is limited. Therefore, the equivalence can be guaranteed in a short time.

In view of such a point, the present invention reduces the area where equivalence guarantee is required by performing the equivalence guarantee by the application program by reducing the area where equivalence guarantee is required by the logic of the application program. Time and resources required for guarantee processing (CPU usage, I / O
The number of times is reduced.

[0015]

The above object is achieved by a duplication volume equivalence assurance processing apparatus configured as described below.

FIG. 1 is a block diagram of the present invention. In the figure, 1 is a volume status management table, 4 is an overall equivalence assurance processing means, 10 is a means of the present invention, 2 is an application-specific equivalence assurance processing target area management table, and 5 is an application-specific equivalence assurance processing. The processing means 3 is an equivalence assurance processing selecting means.

When performing equivalence assurance processing, the equivalence assurance processing selection means 3 knows a volume area in which equivalence can be assured by an application program based on the application-specific equivalence assurance processing target area management table 2, and determines the equivalence of each application. The guarantee processing means 5 is selected and activated. For other volume areas in the volume status management table 1, the equivalence assurance processing selecting means 3 activates the entire equivalence assurance processing means 4 so that the entire equivalence assurance processing by conventional full copying is performed. As described above, since the equivalence assurance processing is performed only on a necessary portion of the volume accessed by the application program, the above-described problems of time reduction and resource saving can be solved.

[0018]

FIG. 2 is a block diagram of an embodiment of the present invention. The figure shows a computer system including a duplicated volume divided into three parts: a processor 20, a nonvolatile medium 21, and a duplicated volume group (29a, 29b, etc.).

A processor 20 controls the system. It consists of three parts corresponding to the means of the present invention. 2
Reference numeral 2 denotes an application-specific equivalency assurance processing unit, which is required to guarantee whether the update processing of the database handled by the application program is collectively updated for a plurality of data in transaction units or none of them is updated. At this time, that is, when the process for guaranteeing the atomicity is being performed, the application program shares the equivalence guarantee process for the database to be updated.

Reference numeral 24 denotes an overall equivalence assurance processing unit provided as a conventional technique. This is because, in the event of an abnormal system stop, it is not possible to know which block on the volume was being written, and it was not possible to identify which part of the volume had lost the equivalence of its contents. The entire volume is copied from one volume to the other to assure the equivalence.

Reference numeral 23 denotes an equivalency assurance processing selection unit that selects and starts the above two equivalence assurance means 22 or 24, and performs an assurance assurance process on a full-duplex volume.
Reference numeral 21 denotes a non-volatile medium which is referred to by the means for guaranteeing the equivalence and stores the operation status of the duplicated volume and the status of the equivalence assurance processing. It is necessary to record the latest state of these duplicated volumes on a non-volatile medium such as a magnetic disk device in order to maintain the state even in an abnormal situation such as a power failure.

Although the details will be described later, reference numeral 25 denotes a duplicated volume status management table. Reference numeral 26 denotes an application-specific equivalence guarantee target area management table. The means of the equivalence assurance processing shown in the above 20 performs the equivalence assurance processing based on the contents shown in these management tables.

Reference numeral 29a denotes a duplex volume A. This is a physically different X-system device 28ax and a Y-system device 28ay
Configured above. That is, the duplicated logical volume on the X-system device and the duplicated logical volume on the Y-system device are the duplicated volume A. A database DATAa is recorded on the duplicated volume A as data. In addition, this DATAa is recorded by the application program on the duplicated volume A as check logs (CKP information a) and data update information a along with the data itself as an update log in order to guarantee the atomicity of the data described above. I have. Therefore, the duplicated volume A is subjected to the equivalence assurance processing by the application program.

The same applies to the configuration of the duplex volume B.
However, the data recorded on this volume is only the data recorded in 27b because the application program using the data does not take any particular action to take an update log. Therefore, the duplicate volume B is a target of the full equivalence assurance processing.

FIG. 3 is a diagram for explaining the relationship between the volume state management and the equivalence assurance processing target for each application. FIG. 3A shows a volume status management table 31, which shows an example of the operation status of a total of four duplicate volumes C and D in addition to the duplicate volumes A and B shown in FIG. Volume status management table 31
Is configured to indicate an access prohibition flag, a duplicated volume status, a copy source volume, and an equivalency assurance processing type corresponding to the duplicated target volume name.

The duplicated volume state includes three states of duplication operation, equivalency assurance processing, and another equivalence assurance processing completion (not shown). The state in which the duplex operation is being performed indicates that the operation is being performed in which the writing is performed twice for the X system and the Y system, and the reading is performed from the volume of one system (for example, the X system). That is, during normal operation of the system, the access method examines this table and writes to both systems as described above. However, as will be described below, when the system is abnormally stopped due to a power cut or the like, there is a state in which the volume can be sequentially accessed in the restart processing, which is different from the duplicated volume state displayed here. As described above, the access prohibition flag is indicated as 1 during the restart processing. FIG. 3A shows that only the volume A has an access prohibition flag of 0 and the others have a value of 1, that is, the equivalence assurance is assured for the volume A in the process of sequentially performing the equivalence assurance processing on the duplexed volume after the system abnormally stops. This indicates that the processing has been completed and the flag indicates that the duplex operation status is displayed again and that access is possible.

As for the volume B, the duplex volume state is in the process of equivalence assurance, that is, the entire copy is in progress.
When the system is in operation, the duplex volume B is operating in a single system, and access can be performed only to the X system indicated by the copy source volume, and data can be read. However,
When the access prohibition flag of the volume B is 1, it indicates that the one-system operation was being performed and the overall equivalence assurance process was being performed concurrently when the system was abnormally stopped.

Another state of the duplex volume state is completion of equivalence assurance processing. This is a state in which equivalency assurance is checked when the system is stopped normally, and duplex operation can be performed at any time when the system is started.

The display of the copy source volume indicates from which system copying is being performed during the equivalence assurance processing. There is no display for volumes A, C and D. Volume B indicates a single-system operation that accesses only the X system.

There are two types of equivalence assurance processing types, one for each application. That is, the present invention distinguishes between a case where the equivalence assurance processing is performed by the application program-specific equivalence assurance processing program and a case where the equivalence assurance processing is performed by the conventional overall equivalence assurance processing.

In addition to the above-described volumes A and B for duplication, volume C is in a duplex operation when the system is abnormally stopped due to power interruption, and the equivalence assurance process uses the entire equivalence assurance process. However, this indicates that the application program is taking an update log, and the equivalence assurance processing for restart uses the application-specific equivalence assurance processing program.

FIG. 3B shows an application-specific equivalence assurance processing target area management table 32. In an application program that guarantees the atomicity (atomicity) of a plurality of data states from the original data update logic, the address of the data being updated can be found from the update log information shown separately. The equivalency assurance processing of a volume can be achieved by starting an equivalence program prepared for each application without copying all blocks of the volume.

Application-specific equivalence guarantee processing target area management table 3
In 2, the application-specific equivalence assurance processing program target volume name indicates a duplicated volume read and written by the application program as described above. As shown in FIG. 3A, target volumes are A and D. The application program name corresponding to each volume is A
Pa and APd, and the start addresses of the application-specific equivalence assurance processing programs are Ta and Td.

FIG. 4 is a flowchart for explaining the operation of the embodiment relating to the equivalence assurance processing at the time of starting the system. When divided according to the configuration of the embodiment of the present invention, FIG.
Is an equivalence assurance processing selection unit, 42 is an application-specific equivalence assurance processing unit, and 43 is an overall equivalence assurance processing unit. The steps will be described below. First, although not shown, the access prohibition flags are set to 1 for all the volumes in the volume status management table of FIG. This is performed to enable sequential access to volumes in the process of performing the equivalence assurance processing of each duplicated volume at system startup.

In step S44, of the volume names in the application-specific equivalence assurance processing target area management table 32 in FIG. 3B, the duplicated volume status is in the duplicate operation in the volume status management table 31 shown in FIG. 3A. Look for any of the following. That is, when the volume status management table for the volume A in the application-specific equivalence assurance processing target area management table 32 is first checked, the duplex volume status is "duplex operation". The control is shifted to S45. Note that FIG.
In (a), the access prohibition flag is 0, which illustrates the state after the equivalence assurance processing for volume A has been completed. However, the access prohibition flag is 1 at the start of the processing.

In step S45, the application program AP for reading and writing the volume A shown in the application-specific equivalence guarantee processing target area management table 32 of FIG.
The control is transferred to the start address of the application-specific equivalence assurance processing program a to start the equivalence assurance processing. When the process is completed, the duplex volume status of the volume A is set to duplex operation in step S46, the access prohibition flag is reset, and both volumes can be accessed as duplex volumes. Similarly, the process returns to step S44 to process the next volume B that meets other conditions. This is repeated until there is no duplicated volume that satisfies the condition of step S44. After completing the equivalency assurance processing for the duplicated volume targeted for the application-specific equivalence assurance processing,
In step S47, the entire equivalence assurance processing of the related art is executed for the remaining duplicated volumes. The full equivalence assurance process often involves copying a large amount of data. In such a case, the volume status management table 31 determines that the copy source is a volume of one of the systems and that the duplex volume status is in the equivalence assurance process. To be read and written by a single operation of only the copy source volume and incorporated into the system. The function of the duplicated volume will be exhibited only when the volume status management table is set to duplicated operation when the complete equivalence assurance processing is completed.

The data update processing of the application program will be described with reference to FIGS. FIG. 5A shows a check point for recording an update log and confirming a process of completing the update data recording when the application program updates a block of data, and what data is to be written to where. 3 shows a format of a table in which storage addresses of data update information are recorded on a non-volatile medium. FIG. 5B is a table showing the format of the data update information described above. The data update information indicates the block address of a series of data to be updated and the start block address on the non-volatile medium storing the value to be set after the update, and the number of consecutive blocks as shown in FIG. It is in the form of just arranging. Further, it is assumed that these pieces of information are arranged on a duplicated volume as shown in FIG.

The operation will be described with reference to the flowchart of the data updating process of the application program shown in FIG. The series of data shown in FIG. 5 is grouped together so that the progress of the processing can be confirmed using checkpoints in order to guarantee and update the atomicity. Step S
In step 61, the state before the start of a series of update processing is recorded by setting 0 to the check point.

In step S62, the value of the data to be written and the address of the write destination are recorded as a data update log. In step S63, 1 is set to the checkpoint information to indicate that the recording of the data update information has been completed.

In step S64, a series of data set in the data update information is updated. In step S65, 2 is set to the checkpoint information to indicate the completion of the data update. (A), (b),
At (c), the range of the arrow indicates how far the data update can be determined based on the value of the checkpoint information.

That is, if the checkpoint information of one system is 2, the range of (a), that is, the data update for the previous time has been completed, and the data and the data update information are both equivalent. However, the checkpoint information may be 0, 1, or 2 depending on the X system and the Y system, and it cannot be said that the equivalence is maintained.

If the checkpoint information of one system is 0, the data update information is being set in the range (b), that is, the update data has been updated last time, and the equivalence is guaranteed. However, the equivalence is not guaranteed for the data update information and the checkpoint information.

If the checkpoint information of one system is 1, the data is being updated in the range of (c) but the completion cannot be confirmed, and the equivalence of the data and the checkpoint data is not guaranteed. This indicates that only the equivalence of the data update information is guaranteed.

A method of performing the equivalence assurance processing by the application program by knowing the above data update status from the checkpoint information will be described with reference to the flowchart of FIG.

In step S71, (a)
In any of (b) and (c), the equivalency of the checkpoints is not guaranteed. Here, the checkpoint information is written from the X system to the Y system to guarantee the equivalence of the checkpoint information. Here, the image may be copied from the Y direction to the X system in the reverse direction.

In steps S72 and S73, the value of the checkpoint information is checked, and the following processing is divided according to the value. As described above, in the case of 0, the data is equivalent, the equivalence state of the data update information is indeterminate, and the routine goes to Step S74. Here, the concept of the copy direction is the same as described above, and the X-system data update information is read and Y
Since the contents do not matter as long as they are copied to the system or are equivalent, a method of initializing the contents, for example, clearing the contents may be used.

Similarly, when the checkpoint information is 1, the data update information is equivalent, but since the equivalence of the data is doubtful, the data update information is read in step S75, and the data is renewed based on the read information. Write in both systems.

If the checkpoint information is 2, the data and the data update information are both guaranteed to be equivalent, so that nothing is done and the equivalence assurance process ends. In the above description, only one group of data is updated by the application program. However, even when there are a plurality of data groups, it is easy to create an equivalence assurance processing program for each application so as to correspond thereto.

[0049]

As is apparent from the above description, according to the present invention, if the equivalence assurance processing can be performed by the logic of the application program in the area of the duplicated volume, the equivalence assurance processing is performed by the application program. It is possible to reduce the volume area required for the performance assurance process and reduce the time and resources (CPU usage, I / O times) required for the overall equivalence assurance process, and especially for continuous operation of a large-scale system. There is an industrial effect of realizing efficiency.

[Brief description of the drawings]

FIG. 1 is a block diagram of the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention.

FIG. 3 is a related explanatory diagram of volume status management and an application-specific equivalence assurance process target;

FIG. 4 is a flowchart illustrating an operation of the embodiment.

Fig. 5 Data update log by application program

FIG. 6 is a flowchart of a data update process of an application program.

FIG. 7 is a flowchart of an equivalence assurance process by an application program.

FIG. 8 is a flowchart of a conventional technique (equivalence guarantee processing at system startup).

[Explanation of symbols]

 1 Volume Status Management Table 2 Application-Specific Equivalence Assurance Processing Target Area Management Table 3 Equivalence Assurance Processing Selection Means 4 Full Equivalence Assurance Processing Means 5 Application-Specific Equivalence Assurance Processing Means

Claims (2)

[Claims] 1. A duplicate volume system in which the same content is written to another volume doubly and a read / write of the other volume can be continued even if a failure occurs in one of the volumes. Is suspected, there is an overall equivalence assurance processing means for completely copying the contents from one volume to the other volume based on the volume status management table indicating the status of the duplicated operation of the volume, and the assurance process of the duplicated volume is performed. An equivalence assurance processing device for a duplicated volume, the application-specific equivalence assurance processing target area management table indicating a volume area in which each application program performs the equivalence assurance processing, and the application-specific equivalence assurance processing for each of the application programs. Based on the target area management table, the Based on the application-specific equivalence assurance processing means that performs the processing, and based on the volume status management table and the application-specific equivalence assurance processing target area management table, as the means of equality assurance for each volume area, the application-specific equivalence assurance processing means or the entire An equivalence assurance processing selecting means for selecting any of the equivalence assurance processing means. 2. A method for guaranteeing the equivalence of a duplicated volume based on a volume status management table for displaying the status of the duplicated operation of a volume, comprising an overall equivalence assurance processing means for completely copying the contents from one volume to the other volume. A management table generating means for generating an application-specific equivalence assurance processing target area management table indicating a volume area in which an application program performs equivalence assurance processing; Based on the application-specific equivalence assurance processing target area management table, the application-specific equivalence assurance processing means for performing the equivalence assurance processing in the target volume area, the volume status management table and the application-specific equivalence assurance processing target area As a means of guaranteeing equivalence for each volume area based on the management table Pre by equivalence assurance processing means or the entire equivalence guarantee equality assurance process selecting means for selecting one of the processing means, a machine-readable recording medium recording a program to function as a.