JP2914181B2 - Disk array device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array device, and more particularly, to a disk array device provided with a cache memory to reduce an overhead time in a data write process.

[0002]

2. Description of the Related Art In recent years, an external storage device connected to a file server or the like has been increasing in capacity with the enlargement of a database. Above all, a disk array device in which disk devices are arranged in an array to achieve high reliability, low cost, and large capacity has been spotlighted. Recently, many related technical documents and patents have been published, and in particular, since the University of California, Berkeley (UCB) in the United States published a paper on array-type disks, it has become publicly known. Became. FIG.
CB's dissertation, "A CASE FOR REDUNDA
NT ARRAYS OF INEXPENSIVE
DISKS (RAID) ", DAVID A. PATT
ERSON, BARTH GIBSON, AND RA
NDY H. KATS, REPORT NO. UCB /
This is a conventional disk array device described as a RAID level 5 in CSD 87/391, DECEMBER 1987.

A conventional disk array device published as a UCB paper will be described below with reference to FIGS. 6 and 7. FIG. FIG. 6 is a configuration diagram of a disk array device, where 1 is a disk array device, 2a, 2b, 2c, and 2d.
Reference numeral denotes a plurality of disk devices constituting the disk array device 1, and reference numeral 3 denotes an exclusive OR operation mechanism for generating new check data in accordance with data update in the disk device. Data is divided and managed in blocks in each disk device. Specifically, a continuous series of data is divided into data blocks 4a, 4b, and 4c, and each data block is divided into disk units 2a, 2b,
2c. The disk device 2d stores a check data block 5 generated by an exclusive OR operation of the data blocks 4a, 4b, and 4c. Thus, the disk devices 2a, 2
One redundancy group 6 is constituted by the data blocks 4a, 4b, 4c on the disk drives 2b, 2c and the check data block 5 on the disk drive 2d. In the figure, ()
Is a path indicating the flow of data. In this way,
In the redundancy group 6, data blocks are stored on the disk devices 2a, 2b, and 2c, and the check data block 5 is stored in the disk device 2d. However, accesses to the disk devices assigned to the check data block 5 are excessively concentrated. In order to prevent this, the check data block 5 is assigned to the disk device 2c in the redundancy group next to the redundancy group 6, and the check data block 5 is assigned to the disk device 2b in the next redundancy group. It is common to shift the device sequentially.

FIG. 7 is a flowchart showing the flow of processing in this conventional example. Next, processing at the time of data writing will be described with reference to the flowcharts of FIGS. Upon receiving the write request, the disk array device 1 receives the _new write data α _new 4 from the request source and transfers it to the exclusive-OR operation mechanism 3 (via the path 1). (S1) The old data α _old 4 corresponding to the new write data is written from the data block 4a on the disk device 2a, and the check data C of the same redundancy group is written from the disk device 2d.
_old 5 is read out (via paths 2 and 3) and transferred to the exclusive OR operation mechanism 3 (S2, S3). Write new data α _new 4 and old data α corresponding to the write data
_{Using the old} 4 and the check data C _old 5, an exclusive OR operation is performed on the exclusive OR operation mechanism 3 of the disk array device 1 to generate new check data C _new5 . (S4) Finally, new data α _new 4 and new check data C _new 5
To the corresponding data blocks 4a and check data blocks 5 on the disk devices 2a and 2d (pass 4,
(Via 5). (S5, S6) In this way, data in an arbitrary data block constituting the redundancy group 6 can be generated by performing an exclusive OR operation on the check data block and another data block in the same redundancy group. Therefore, even if the data in any one disk device in the redundancy group is corrupted, the data in the other redundancy device in the same redundancy group and the check data are used to reproduce the corrupted data and realize the failure recovery process. can do.

FIG. 8 is a system diagram showing an example of a disk array device disclosed in Japanese Patent Laid-Open No. 5-324206, and FIG. 9 is a schematic diagram thereof.

[0006] When the microprocessor 101 receives a read request from the host computer via the host interface 102, the microprocessor 101 corresponds to the disk device dr3 storing the data C specified by the read request and corresponding to the data. For the disk device dr5 in which the check data P is stored, the corresponding data C,
Request reading of check data P.

[0007] The microprocessor 101 transmits the corresponding data C and check data P read from the disk devices dr3 and dr5 to the drive controller C, respectively.
When input via NTL3 and CNTL5, those data are stored in cache memory 104 (via paths 1 and 2).
After storing, only the data C is transferred to the host computer via the host interface 102.

Next, when the microprocessor 101 receives a request to write the update data C ′ for the data C via the host interface 102, the microprocessor 101 receives the request for updating the data C stored in the cache memory 104.
Then, the check data P is transferred to the parity generation unit 103 (via the paths 3 and 4). Further, the update data C ′ is also transferred to the parity generation unit 103 (via the path 5).

The parity generator 103 performs an exclusive OR operation on the old data C, the old check data P, and the update data C ′ to generate new check data P ′.

The microprocessor 101 receives the new check data P 'from the parity generation unit 103 and sends the updated data C' and the new check data P 'to the drive controllers CNTL3 and CNTL, respectively.
5, the disk drive dr3 and the disk drive dr5
(Via passes 6 and 7).

[0011]

The UCB paper "AC
as for Redundant Arrayso
f Inexpensive Disks (RAI
D), the old data α _old 4 and the old check data C _old 5 are written to the disk devices 2 a and 2 d to generate check data when writing data to the disk array device 1. , Read the data into the exclusive OR operation mechanism 3 and write the new data α _new 4, the old data α _old 4, and the old check data C
_old 5 and performs an exclusive-OR operation is, after generating a new check data C _{new new} 5, for writing new check data C _{new new} 5 and the new data alpha _{ne w} 4 disk units 2a, to 2d, the disk during data writing There has been a problem that access to the device occurs four times and the data writing process becomes slow.

In the disk array device disclosed in Japanese Patent Application Laid-Open No. 5-324206, after reading the data C shown in FIG. 9 to the host and writing the data to the disk device, data other than the data C, for example, data D When a read request occurs, the check data P 'exists in the cache memory, but there is a problem that the data D must be read from the disk device dr4.

When the cache computer 104 is already full of previously referenced data and check data when a new data read request is issued from the host computer, the cache memory 104 stores which data and which check data are cached. No consideration is given to whether the data is deleted from the memory 104 and new data is newly read into the cache memory. Depending on the deleted data and the check data, the efficiency of the reading and writing processing for the disk array device is reduced, and the disk access speed is reduced. However, there has been a problem that is greatly reduced.

SUMMARY OF THE INVENTION The present invention has been made to solve the above-mentioned problems, and a disk array device which improves the input / output processing speed by reducing the number of accesses to the disk device in the disk array device. The purpose is to provide.

[0015]

According to a first aspect of the present invention, there is provided a disk array device comprising: a plurality of data disk devices; and at least one or more check data disk devices.
A temporary storage unit for temporarily storing data in a process of reading data stored in the data disk device and a process of writing to the data disk device, and controlling input / output of data between the temporary storage device and the disk device Input / output control means, and check data for generating new check data corresponding to the update data based on the update data, the old data to be updated, and the check data corresponding to the old data in the data update process for the data disk device In the disk array device comprising the generation operation mechanism, the input / output control means controls all of the same redundant group consisting of a data block containing read data and a corresponding check data block at the time of data read processing from the disk array device. Disk data The new check data is generated based on the old data stored in the temporary storage means and the check data corresponding to the old data during the data update processing for the disk array device. Things.

A disk array device according to a second aspect of the present invention
In the disk array device according to the first invention, the input / output control means reads out from each disk device a redundant group including read data and a next redundant group following the redundant group in the process of reading data from the disk array device. is there.

A disk array device according to a third aspect of the present invention
In the disk array device according to the first or second aspect of the present invention, when the data read into the temporary storage means is full as a data management method on the temporary storage means, the data which has not been most recently referred to. , And a method of sequentially replacing the new data with new data.

[0018]

In the disk array device according to the present invention, a redundant group consisting of write data to the disk device and check data generated in correspondence with the write data is configured as a unit. To temporarily store the read result in the cache memory, and in the subsequent check data generation process in the disk write process, the old data for the necessary write data and the check data corresponding to the old data are extracted from the cache memory. It was made.

Further, in the disk array device according to the present invention, in the process of reading from the disk device, two groups, a redundant group including read request data and a next redundant group following the redundant group, are simultaneously read and stored in the cache memory. It is intended to be stored.

Further, according to the disk array apparatus of the present invention, when data read into the temporary storage means is full as a data management method on the temporary storage means, the data which has not been referred to most recently can be used. The new data is replaced in order.

[0021]

【Example】

Embodiment 1 FIG. Hereinafter, a first embodiment of the present invention will be described with reference to FIGS. FIG. 1 is a block diagram showing an embodiment of the present invention. In FIG.
2a, 2b, 2c, 2d are disk devices, and 4a,
4b and 4c are data blocks constituting the redundancy group 6 on the disk devices 2a, 2b and 2c, respectively.
Is a check data block in the redundancy group 6. Numeral 3 denotes an exclusive OR operation mechanism, numeral 7 denotes a cache memory, and numeral 6 'denotes a state in which the redundant group 6 is loaded on the cache memory 7. Also,
Reference numeral 8 denotes a cache memory management mechanism for managing the cache memory, and reference numeral 9 denotes input / output control means for controlling data input / output with the disk device. In the figure, (
) Is a path showing the flow of data and control

FIG. 2 is a flowchart showing the flow of a process of reading data from the disk array device in this embodiment, and FIG. 3 is a flowchart showing a process of writing data.

First, regarding the processing at the time of reading data,
A description will be given based on the flowcharts of FIGS.
When the disk array device 1 configured as shown in FIG. 1 receives a data read request (SR1), the cache memory management mechanism 8 determines that the redundancy group 6 ′ including the requested data α _new 4 has already been stored in the cache memory 7. It is determined whether it exists (SR2). Here, the corresponding request data α _new 4 is the data block 4 of the disk device 2a.
Assume that it exists in a. If the redundant group 6 including the request data α _new 4 does not exist in the cache memory 7 and an empty area remains in the cache memory 7 (SR3), the input / output control unit 9 sets the disk device 2a, 2b, 2c, the corresponding request data α _{ne w} 4 from 2d
, All data blocks 4a, 4b,
4c, the check data block 5 is read out (via pass 1) into a free area of the cache memory 7 (SR5).
When the cache memory 7 is full, the cache memory management mechanism 8 searches the redundant group data in the cache memory 7 for the head area of the redundant group having the longest time since the last reference. Yes (SR
4). As described above, the loading addresses of all data constituting the redundancy group 6 have been determined.
_The redundant group data 6 including the _new 4 is stored (overwritten) in the searched location on the cache memory 7 (S
R5). Next, the corresponding request data α _new
4 (via path 2) (SR6).

On the other hand, if the redundant group 6 'including the requested data α _new 4 already exists on the cache memory 7 (SR2), the data block 4a on the cache memory 7 is sent to the request source. Request data α _{new from}
4 directly (via path 2). (SR6).

The cache memory management mechanism 8 controls the data time management pointer to point to the last location where the data of the redundancy group 6 'is stored on the cache memory 7 (SR7), and notifies the end of reading. Is sent to the request source (SR8), thereby ending a series of processing.

Here, for example, in the case where the data read request has a content that extends over the data blocks 4b and 4c, if there is no redundant group 6 'including the relevant data in the cache memory 7, the disk Applicable data block 4b from devices 2a, 2b, 2c, 2d
Data blocks 4a, 4a,
4b and 4c and the check data block 5,
After storing in the cache memory 7, the corresponding request data 4b and 4c are transferred to the request source.

Next, the processing at the time of writing the data 7 will be described with reference to the flowcharts of FIGS.
The disk array device 1 receives a data write request (SW1) and receives write data (SW2).
The disk array device 1 transfers the write data received from the request source to the exclusive OR operation mechanism 3 (via the path 3) (SW3). Here, when the redundant group data 6 'including the write data exists in the cache memory 7 (SW4), the old data α _old 4 and the old check data C _old 5 in the cache memory 7 are exclusive-logically combined. Transfer (via paths 4 and 5) to sum operation mechanism 3 (SW
8). In the exclusive OR operation mechanism 3, the new data α _new 4 transferred from the request source, the old data α _old 4 transferred from the cache memory 7, and the old check data C
performs an exclusive OR operation with the _old 5, to generate the new check data C _new 5 (SW9). The disk array device 1 writes the new data α _new 4 and the generated new check data C _new 5 to the respective disk devices 2a and 2d (via the paths 7 and 8) (SW10). Further, writing is performed on the corresponding data block in the redundancy group 6 'on the cache memory 7 (via the paths 9, 10), and the contents of the cache memory are updated. (SW1
1) Next, the redundant group 6 'is inserted at the end indicated by the pointer managed by the cache memory management mechanism 8 (S
W12). Finally, a write completion notification is sent to the request source (SW13), and a series of write processing is completed.

On the other hand, if the redundant group data including the write request data does not exist in the cache memory 7 (SW4), the cache memory management mechanism 8 searches for a free area in the cache memory 7 (SW5) and finds a free area. For example, the disk device 2
a, 2b, 2c, and 2d, the redundant group data including the corresponding data is read and stored in the cache memory (S
W7). If there is no free area in the cache memory 7 (SW5), the leading redundant group indicated by the pointer managed by the cache memory management mechanism 8 is extracted (SW6), and the disk devices 2a, 2b, 2c,
The redundant group data read from 2d is stored (S
W7). Subsequent processing is the same as when the redundant group data including the relevant data exists in the cache memory 7 (SW8 to SW13).

Embodiment 2 FIG. A second embodiment of the present invention will be described with reference to FIGS. FIG. 4 is a diagram showing an embodiment of the present invention, wherein 6a and 6b show continuous redundant groups on the disk device, and 6a 'and 6a'
b 'shows how these redundant groups are loaded on the cache memory. Other components are the same as those described in the first embodiment. FIG. 5 is a flowchart showing the flow of the data reading process in the disk array device of the present embodiment.

Next, the operation will be described with reference to the flowchart of FIG. When the disk array device 1 receives the data read request (SR1), it is determined whether or not the redundancy group 6a 'including the requested data α _new 4 already exists on the cache memory 7 (SR).
2). Redundancy group 6 including requested data α _new 4
a 'does not exist in the cache memory 7 and a free area remains in the cache memory 7 (SR3)
The input / output control means 9 includes the disk devices 2a, 2b, 2
From c and 2d, all data blocks 4a, 4b and 4c of the redundant group 6a including the corresponding request data α _new 4, and all the data blocks 4d and 4e of the check data block 5a and the redundant group 6b continuous with the redundant group 6a.
4f and the check data block 5b are read (via pass 1) into a free area of the cache memory 7 (SR
5). When the cache memory 7 is full, the cache memory management mechanism 8 is used to search for two redundant group data that have passed the longest time since they were last referenced from among the redundant group data in the cache memory 7. (SR4). As described above, the loading addresses of the new redundant groups 6a and 6b are determined, and these two continuous redundant group data including the requested data α _new 4 are stored (overwritten) in the above-described searched location on the cache memory 7. (SR5). Then, the corresponding request data α _new 4 is transferred (via path 2) to the request source (SR6). Next, the redundant groups 6a 'and 6b' on the cache memory 7 are sequentially inserted into the last position of the pointer indicated by the cache memory management mechanism 8 (SR7), and a read end notification is sent to the request source (SR).
10) completes a series of processes.

On the other hand, if the redundant group 6a 'including the requested data α _new 4 already exists on the cache memory 7 (SR2), the redundant group 6a on the cache memory 7 is notified to the requestor. 'To the corresponding data α
_{transfer new} 4 (via path 2) (SR8) Next, the cache memory management mechanism 8 inserts the redundancy group 6a 'in the cache memory 7 into the last position indicated by the time management pointer of the data (SR9), A notification of the end of reading is sent to the request source (SR10), and a series of processing is completed.

The writing process to the disk array device is the same as in the first embodiment.

[0033]

Since the present invention is configured as described above, in the check data generation processing at the time of writing data to the disk array device, old data and old check data corresponding to the write data are read from the disk device. Since the data is read out from the temporary storage means without any problem, it is possible to improve the disk access processing performance.

In the data reading process, not only the read data and the corresponding check data but also all data constituting the redundant group are simultaneously read into the temporary storage means. In addition, high-speed access can be made to other data accesses in the same redundant group that is continuous with the above.

In the data reading process, the data constituting the next redundant group adjacent to the redundant group containing the data is also read into the temporary storage means at the same time. Becomes possible.

Further, when the temporary storage means is already full of data which has been read before, the data block which has not been referred to recently is replaced with a new data block in order from the most recently read. In this process, useless operation such as sweeping out a data block having a high access probability can be prevented as much as possible.

[Brief description of the drawings]

FIG. 1 is a configuration diagram showing a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a data reading operation of the disk array device according to the first embodiment of the present invention.

FIG. 3 is a flowchart illustrating a data write operation of the disk array device according to the first embodiment of the present invention.

FIG. 4 is a configuration diagram showing a second embodiment of the present invention.

FIG. 5 is a flowchart illustrating a data read operation of the disk array device according to the second embodiment of the present invention.

FIG. 6 is a configuration diagram of a disk array device in a conventional example.

FIG. 7 is a flowchart showing the operation of the conventional disk array device.

FIG. 8 is a block diagram showing a system configuration of a conventional disk array device.

FIG. 9 is a schematic diagram showing the operation of a conventional disk array device.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Disk array device 2 Disk device 3 Exclusive OR operation mechanism 4a, 4b, 4c, 4d, 4e, 4f Data block 5 Check data block 6 Redundancy group 7 Cache memory 8 Cache memory management mechanism 9 Input / output control means

────────────────────────────────────────────────── ─── Continuation of the front page (56) References JP-A-6-119120 (JP, A) JP-A-6-95810 (JP, A) JP-A-6-95968 (JP, A) (58) Field (Int.Cl. ⁶ , DB name) G06F 3/06 302 G06F 3/06 305 G06F 3/06 540 G06F 12/08 320 G06F 12/16 320

Claims (3)

(57) [Claims] 1. A plurality of data disk devices and at least one or more check data disk devices, and data stored in the data disk device are read and data are written to the data disk device. Comprising: a temporary storage unit for temporarily storing data; an input / output control unit for controlling input / output of data between the temporary storage unit and the disk device; and a check data generation operation mechanism for generating new check data. In the array device, the input / output control unit reads out, from the disk device, all data of the same redundancy group composed of a data block including the read data and a corresponding check data block in a process of reading data from the disk array device Stored in the temporary storage means, The check data generation operation mechanism responds to the update data based on the update data, the old data to be updated stored in the temporary storage means, and the check data corresponding to the old data in the data update processing for the disk array device. A disk array device characterized by generating new check data. 2. The data processing method according to claim 1, wherein said input / output control means reads a redundant group including the read data and a next redundant group following the redundant group from each disk device in a process of reading data from the disk array device. The disk array device according to claim 1. 3. The input / output control unit includes a management unit for the temporary storage, wherein the temporary storage management unit is not most recently referred to when the data read into the temporary storage unit is full. 3. A device according to claim 1, further comprising means for sequentially replacing new data with new data.
Item 14. The disk array device according to Item 1.