JP2003173240A - Storage device system and its data backup method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To reduce memory capacity required for control, and to shorten time required for re-synchronization to doubling in a method for controlling difference data between both volumes after transferring the doubling volume for constituting a pair to a separate state from a doubling state. <P>SOLUTION: A storage area on the volume is controlled by LU securing block information 200, a block control table 210, and a bit map area 220. One bit of the bit map area 220 is made to correspond to the storage area (for example, 1 MB) equal to the maximum transfer length when copying data between the volumes to control the existence of the difference data of the respective storage areas on the volume. The bit map area 220 is divided into a plurality of blocks, and can make access to a specific bit allocated to the respective one bit corresponding storage areas of the respective volumes via a block number. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a data backup system for a storage device system, and more particularly to a system for backing up data by duplicating volumes.

[0002]

2. Description of the Related Art A user volume in a disk system is backed up in order to protect the data from system failure, human error, disaster and the like. For example, in the method disclosed in Japanese Patent Laid-Open No. 11-153386, as shown in FIG. 7, a copy of the user volume 70 that stores user data and is used for online work is stored in the secondary volume 80. The duplexed state of the user volume 70 and the secondary volume 80 is called a pair. Tape device 6
When the data is not backed up to 0, the data is redundantly stored in the user volume 70 and the secondary volume 80. When backing up data to the tape device 60, the host computer 10
Issues a command for instructing the pair separation to the disk system 20 and also issues a copy command to the secondary volume 80 to copy all the data stored in the separated volume to the tape device 60. On the other hand, when the disk system 20 receives the pair separation instruction, it separates both volumes of the specified pair.
When the copy command is received, the designated block is data-copied from the secondary volume 80 to the tape device 60. Secondary volume 80 when pair is separated
Since the above data is frozen, access to the user volume 70 and data copy processing from the secondary volume 80 can be performed in parallel after the pair separation.

Even while the copy from the secondary volume 80 to the tape device 60 is being executed, the access from the host computer 10 to the user volume 70 can be continued. When the data on the user volume 70 is updated, it is recorded that the updated block 71 has not been reflected in the secondary volume 80. When the data copy to the tape device 60 is completed, the host computer 10 issues a resynchronization command to return the user volume 70 and the secondary volume 80 to the duplicated state again. At this time, the update data of the user volume 70 is copied to be reflected in the secondary volume 80.

[0004]

When data is updated to the user volume 70 in the separated state of the pair, it is recorded in the secondary volume 80 that it has not been reflected. Since the unit to record at this time is one block (512B) per bit, the volume capacity is 2TB.
Then, a memory capacity of 1 GB is required. This is 1
It is a memory required for each pair, and if the number of pairs is increased by n times, the required memory capacity becomes 1 GB × n, and a huge amount of memory is required. However, conversely, the smaller the size per bit, the shorter the resynchronization time when the data difference between both volumes is small. Therefore, considering the resynchronization time, the smaller the storage unit per bit is, the better. Become.

In view of the above, an object of the present invention is to reduce the memory capacity necessary for recording unreflected data in the secondary volume in the pair separation state, and to shorten the resynchronization time associated therewith.

[0006]

According to the present invention, a first volume and a second volume constitute a pair for data backup, and the first volume and the second volume are transferred from a duplexed state to a separated state. In the data backup method of the storage device system that manages the storage area where data has been written to the volume, it is equal to the maximum transfer length that can be transferred once when copying data from the first volume to the second volume. The storage unit corresponding to the storage area on the first volume is set on the storage device as the management unit for storage area management in which data has been written, and the management unit corresponding to the storage area in which data has been written is specified and specified. It is characterized by a data backup method that sets a flag indicating that there is a write in the selected management unit.

[0007]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is a diagram showing the configuration of a system related to the present invention. The storage system 20 is connected to the host computer 10. The storage device system 20 is
In terms of hardware, it is configured by components such as a CPU, a memory, a cache, and a disk storage device, and operates by a control program stored in the memory. The disk storage device viewed from the host computer 10 is recognized as a logical volume. In FIG. 1, illustration of these constituent elements is omitted, and the MRCF-Lite function (Mul) is a function of a control program and is related to the present invention.
triple RAID Coupling Feature
re-Lite) 30 is shown. MRCF-Li
The te function 30 is activated by a command from the host computer 10. The primary volume (P-Vol) and the secondary volume (S-Vol) form a mirror pair for data backup. The MRCF-Lite function 30 manages P-Vol and S-Vol. The volume will be described as an LU (logical unit) hereinafter.

The MRCF-Lite function 30 includes a duplication function (coupling) 31 and a separation function (split).
32 and a resynchronization function (resync) 33. The duplication function 31 copies the data of the primary LU 40 to the secondary LU 50 to copy the primary LU 40 and the secondary LU 50.
Duplicate 50 as a mirror pair. When the data is written to the primary LU 40, the duplication function 31 also writes the same data to the secondary LU 50 to match the data of the primary LU and the secondary LU.

The separating function 32 is implemented by the host computer 10.
When the mirror pair separation is instructed by, the data duplication is stopped at that point. Data is written to the primary LU 40 WR
Even if it is ITE, it is not reflected in the sub-LU 50, and conversely the sub-LU
Even if the data is written in 50, it is not reflected in the primary LU 40.

The resynchronization function 33 is provided in the host computer 1
When the mirror pair resynchronization is instructed by 0, data is copied from the primary LU 40 to the secondary LU 50 or from the secondary LU 50 to the primary LU 40 according to the instruction, and the separated primary LU 4
The data of 0 and the data of the sub-LU 50 are matched again.

FIG. 2 shows an L in the storage system 20.
It is a state transition diagram which shows the state of a pair of U, and the conditions which each state changes. The pair is released "SMPL" 10
When a command for instructing duplexing is received from the host computer 10 at 0, the status "COPY" is being executed during the entire volume copy (initial copy) from the primary LU to the secondary LU.
(PD) ”101. When copying is completed, the primary LU
"PAIR" 10 that the data of the sub-LU and the data of the sub-LU match.
2, the duplication is completed and both LUs become a mirror pair. Host computer 10 in "PAIR" 102 state
When the command to separate the pair is received from the device, the status changes to "PSUS" 103 and the data is transferred to the primary LU 40.
Even if RITE is performed, it is not reflected in the sub-LU 50, and the sub-LU 5
Unless there is a WRITE of data for 0, the data of the primary LU 40 at the time of separation is held. "PSUS" 103
When a command to resynchronize from the primary LU to the secondary LU is received from the host computer 10 in the state, the copy state from the primary LU to the secondary LU becomes “COPY (RS)” 104. Host computer 10 in "PSUS" 103 state
When a command to resynchronize from the secondary LU to the primary LU is received, the status of copying from the secondary LU to the primary LU "COP
Y (RS-R) ”105. When the copy for resynchronization is completed, the state returns to the“ PAIR ”102 state and the duplication is completed. When the pair state is other than“ SMPL ”100, the suspend instruction is issued from the host computer 10. When it is received, the suspended state becomes "PSUE" 106. Also, when a failure occurs during the initial copy or resynchronization, it becomes "PSUE" 106. "COPY (RS)" 1
When the suspension in 04 or “COPY (RS-R)” 105 is canceled or the failure is recovered, the original copy state is restored. When the command for canceling the pair is received from the host computer 10, the process returns to "SMPL" 100.

FIG. 3 is a diagram showing the data structure of the LU reserved block information 200 stored in the memory of the storage system 20, the block management table 210, and the bitmap area 220 provided in the cache area. The bitmap area 220 is divided into a plurality of blocks, and each block has a storage capacity of 2 KB. Each block corresponds to one of the block numbers starting from 0. Since 1 bit in a block corresponds to a storage area of 1 MB of LU, 16 blocks of LU
The storage area for GB is managed. Here, 1 MB corresponds to one maximum data transfer length for LU-LU copy due to hardware restrictions. That is, 1 bit in the bitmap area 220 corresponds to the maximum transfer length.
The bitmap area 220 is a primary L that constitutes an LU pair.
Bitmap information is stored for each U and sub-LU. Each bit is called a difference bit. Initial state and primary LU
And the contents of the storage area corresponding to the difference bit of the sub-LU match, the difference bit is set to the value “0” (off). "P
When information is written to the differential bit corresponding storage area of the primary LU or the secondary LU in the SUS "103 state, the corresponding differential bit is set to the value" 1 "(on).

The block management table 210 has a block number 0 for each block in the bitmap area 220.
The in-use flag 211 and the block start address 212 are stored in this order. The busy flag 211 is a flag indicating whether the block is in use. The block start address 212 sets the start address of the block in the bitmap area 220. The LU reservation block information 200 sets the block number of the allocated block in the bitmap area 220 for each LU. Although not shown, the LUs of the primary LU and secondary LU that make up the LU pair
A set of numbers is stored in the memory as control information.

A duplication function 31 of the storage system 20
When receiving a duplication instruction command from the host computer 10, refers to the block management table 210 for each of the primary LU and the secondary LU that make up the LU pair, and determines the block whose busy flag 211 indicates unused to the capacity of the LU. According to the above, the necessary amount is allocated and the busy flag 211 corresponding to the allocated block is changed to the busy state. In addition, the block number of the block assigned to each LU is L
Register in the U reservation block information 200. Since the required amount of bitmap area 220 is secured in advance,
There is no shortage of blocks to allocate. Upon receiving the command for canceling the pair from the host computer 10, the duplication function 31 returns all the in-use flags 211 of the corresponding block of the block management table 210 to the unused block for the corresponding LU pair. To release. Further, the block number information of the corresponding LU in the LU reserved block information 200 is deleted.

The maximum number of blocks that can be allocated per LU is 1
By using 28 blocks, it is possible to manage the bitmap of the differential bits corresponding to the LU having the maximum storage capacity of 2 TB. Therefore, the memory capacity required to manage the difference information of an LU pair composed of LUs having a capacity of 2 TB is “bitmap area 220 (2 KB × 128 blocks × 2 LU) + LU reserved block information 200 (several bytes × 128 Block × 2 LU) + block management table 210 (several bytes × 128 blocks × 2 LU) = about 5
13 KB ". When managing multiple LU pairs,
"Memory required for 1 LU pair x number of pairs".

When the data writing to the secondary LU is frozen in the separated state, the LU reserved block information 20
0, the block management table 210, and the sub-LU for the bitmap area 220 are unnecessary, so the storage capacity required for them is about half the above.

If the storage capacity corresponding to the difference bit, which is a management unit, is reduced, more memory is required for one LU pair, but the resynchronization time is shortened when the difference is small.
Conversely, if the storage capacity corresponding to the difference bit is increased, it becomes 1
Although the memory required for the LU pair is small, the amount of copying is large when the difference is small, and the resynchronization time is long.

FIG. 4 is a flowchart showing the flow of the differential bit ON setting process performed by the separation function 32 of the storage system 20. The separation function 32 receives the WRITE command from the host computer 10 for the primary LU or the sub-LU which constitutes the LU pair, and when the LU pair is in the “PSUS” 103 state (step 301YE
S), referring to the bitmap area 220 via the LU reserved block information 200 and the block management table 210, the position of the differential bit in the bitmap area 220 corresponding to the write area of the LU is calculated (step 3).
02), information of 1 byte including the corresponding difference bit is acquired from the bitmap area 220 on the memory (step 303). Next, the value of the relevant difference bit is set to ON (step 304), and 1 byte of information including the ON-set difference bit is stored in the original position in the bitmap area 220 (step 305).

FIG. 5 is a flow chart showing the flow of resynchronization processing performed by the resynchronization function 33. Resync function 33
Searches for the specified LU pair in order via the LU allocation block information 200 and the block management table 210 for the difference bits for which the corresponding value of the primary LU or the secondary LU in the bitmap area 220 is ON (step 401). If there is a corresponding difference bit (step 4)
02), the information of the storage area corresponding to the difference bit is copied from the primary LU to the secondary LU or from the secondary LU to the primary LU according to the resynchronization instruction (step 403). The storage capacity to be copied at this time is the maximum transfer length of LU-LU copy (1
MB). From the host computer 10 READ /
When the copying is interrupted due to a reason such as receiving a WRITE request (step 404 NO), step 40
Returning to 3, the differential copy is repeated until the copy of the differential bit corresponding storage area is completed. When the copy of the differential bit corresponding storage area is completed (step 404 YES), the differential bit is turned off, the process returns to step 401, and the search for the remaining differential bits is continued. When the difference bit of ON is not found and the search for the LU pair is completed (NO in step 402), the process is ended.

FIG. 6 is a diagram for explaining the details of the processing for searching and discriminating the difference bit in steps 401 and 402. For example, the block number of the block allocated to the primary LU is acquired from the LU reserved block information 202, and the 1-byte 221 of the bitmap included in the block of block number 0 is acquired. Similarly, the block number of the block assigned to the sub-LU is used as the LU reserved block information 2
03, and one byte 222 of the bitmap included in block number 6 is acquired. Byte 223 is obtained by taking the logical sum (OR) of the difference bits for 1 byte 221 and 1 byte 222. The bit whose value is ON in the byte 223 is the corresponding bit with a difference. The resynchronization function 33 copies the information in the storage area corresponding to the difference bit with a difference according to the resynchronization instruction. When the copy is completed, the corresponding difference bit, which was the primary LU or the secondary LU, is turned off. Note that it is also possible to search for the difference bit only for the primary LU.

When the resynchronization processing is completed for the LU pair in this way, the primary LU and the sub-L of the LU pair are completed.
All the difference bits corresponding to U are turned off. The duplication function 31 sets the pair status of the LU pair to "PAIR" 102, and duplication is completed.

When copying information from LU to LU, 1
A hardware overhead time is added to the copy time for each data transfer. Therefore, if the differential bit correspondence storage area is set to the maximum transfer length, there is a trade-off between the memory capacity required for managing differential data and the resynchronization required time.
Optimized. If the differential bit corresponding storage area is made longer than the maximum transfer length, the memory capacity is reduced, but an extra overhead time is added, which is disadvantageous. If the differential bit corresponding storage area is made smaller than the maximum transfer length, the memory capacity increases and an extra overhead may be added, which is disadvantageous.

[0024]

As described above, according to the present invention, when managing the differential data of both volumes in the separated state of both volumes forming a pair, by setting the management unit to the maximum transfer length, the difference The memory capacity required for management can be reduced, and the time required for resynchronization can be shortened.

[Brief description of drawings]

FIG. 1 is a diagram showing a configuration of a system related to the present invention.

FIG. 2 is a state transition diagram showing LU pair states and conditions under which each state transitions.

FIG. 3 is a diagram showing a format example of management information for difference management.

FIG. 4 is a flowchart showing a flow of differential bit ON setting processing according to the embodiment.

FIG. 5 is a flowchart showing a flow of resynchronization processing according to the embodiment.

FIG. 6 is a diagram illustrating a process of searching and determining a difference bit according to the embodiment.

FIG. 7 is a diagram illustrating a conventional data backup method.

[Explanation of symbols]

10 ... Host computer, 20 ... Storage system, 30 ... MRCF-Lite function, 31 ... Duplication function, 32 ... Separation function, 33 ... Resynchronization function, 40 ... Primary LU, 50 ... Sub LU, 200 ... LU reserved block information, 210 ... Block management table, 220 ... Bitmap area.

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI theme code (reference) G06F 12/16 310 G06F 12/16 310M (72) Inventor Takaaki Miura 6-chome, Onoue-cho, Naka-ku, Yokohama-shi, Kanagawa 81 Hitachi Software Engineering Co., Ltd. In-house (72) Inventor Hiroshi Izumo 6-81 Onoue-cho, Naka-ku, Yokohama-shi, Kanagawa Hitachi Software Engineering Co., Ltd. In-house (72) Inventor Masaaki Kobayashi 322-2 Nakazato, Odawara-shi, Kanagawa No. F-term in the RAID Systems Division, Hitachi, Ltd. (reference) 5B018 GA04 HA04 KA11 MA12 5B065 BA01 CA12 CA16 CC03 CC08 CC10 EA31 EA33 5B082 DC09

Claims (3)

[Claims] 1. A first volume and a second volume constitute a pair for data backup, and data writing to the first volume is performed after the both volumes have transitioned from a duplexed state to a separated state. In the data backup method for a storage device system that manages a storage area that has a problem, the first backup data is equal to the maximum transfer length that can be transferred at one time when data is copied from the first volume to the second volume. A storage unit corresponding to a storage area on a volume is set on the storage device as a management unit for managing the storage area in which the data has been written, and the management unit corresponding to the storage area in which the data has been written is specified and specified. A data backup method, characterized in that a flag indicating that there is writing is set in the management unit that has been created. 2. When the resynchronization instruction is received for both the separated volumes forming the pair, the management unit in which the flag indicating the writing is set for the first volume is searched. The data backup method according to claim 1, wherein data is copied from the first volume to the second volume in a storage area on the first volume corresponding to the management unit of the search result. 3. A first volume and a second volume form a pair for data backup, and both the volumes are written to the first volume after transition from a duplexed state to a separated state. In a storage device system having means for managing a storage area that has a problem, the first volume is equal to the maximum transfer length that can be transferred at one time when data is copied from the first volume to the second volume. Means for setting a storage unit corresponding to the upper storage area on the storage device as a management unit for storage area management in which the data has been written, and means for specifying the management unit corresponding to the storage area in which the data has been written And a means for setting a flag indicating that there is writing in the specified management unit.