JPH09237224A - Disk array device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To execute an efficient operation in either of the cases where a host system is able or vnable to execute multi-access. SOLUTION: When a request from the host system is multi-acess, a cache memory control part 2 sets the unit of access to respective magnetic disk devices 3 and 4 to be one block being the unit of distribution arrangement to the respective magnetic disk devices 3 and 4 and stores data block 8 (9) including the data in a cache memory 1 with a read request to either magnetic disk device 3 (4). If the read request to either magnetic disk device 3 (4) is given when the request from the host system is not multi-acess, the data block 8 (9) including data and a block following it are stored in the cache memory 1.

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 specifically, when a read request is issued from a host, data of an address near a read target address is stored in a cache memory. The present invention relates to a disk array device capable of speeding up response to a read request from a host.

[0002]

2. Description of the Related Art A disk array device is a file device incorporating a plurality of magnetic disk devices using the "RAID architecture" devised at the University of California, Berkeley. Among the "RAID architecture", RAID level 0, level 4, level 5
In the type called so called, data is distributed in a block unit in the built-in magnetic disk devices, and high throughput can be obtained by accessing these data in parallel. Further, in RAID levels 4 and 5, since redundant data is stored in the area of one built-in magnetic disk device, it is possible to continue processing even if one device fails, and high reliability is achieved. Is featured.
Hereinafter, the RAID levels 0, 4, and 5 will be simply referred to as disk array devices. It is said that this disk array device can realize high reliability and high throughput.

FIG. 2 shows a typical configuration of this disk array device. In this figure, 31 is a control unit of the disk array device, 32 to 35 are magnetic disk devices incorporated therein, and 36 to 39 are disk interfaces for connecting the control unit 31 and the magnetic disk devices 32 to 35, respectively. The disk array device is connected to the host system by a host interface (host I / F) 40. Further, reference numeral 41 is a cache memory built in the disk array device, and is composed of a processor 42 which controls this cache memory. The number of built-in magnetic disk devices is not specified in the RAID architecture, but the four-device configuration shown in FIG. 2 will be described below as an example.

The disk array device generally behaves as if it were one magnetic disk device to the host system, but the disk address of this device as seen from the host system (data seen from the host system). Data addresses 43 to 4 consecutive on
6 are actually arranged on individual built-in magnetic disk devices.

FIG. 3 is an explanatory diagram of an example of arrangement of data blocks in the disk array device. 43 to 46 in the figure
Are data blocks, for example, continuous addresses d0, d1, of data viewed from the host system.
Actually, d2 and d3 are the magnetic disk devices 43 and 4,
4, 45, and 46, respectively. Further, the data sizes of the data blocks 43 to 46 are the same,
It is an integral multiple of the sector capacity of the built-in magnetic disk device.

Next, a case where a read request is issued from the host system to this disk array device will be described. If the host system is multi-accessible,
Multi-access may not be possible. Here, a multi-accessible host system is to issue multiple requests at the same time by issuing a certain access request to the magnetic disk device and then issuing the next different request without waiting for the response to the request. Further, it is possible to return the responses in the order in which the disk array device can execute them, regardless of the request order. On the other hand, a host system that is not multi-accessible is a system that performs access sequentially.

FIG. 4 is an explanatory diagram of read access in the case of multi-access, in which a read request from a multi-accessible host system is sent to all the magnetic disk devices 43 to 46 in the disk array device. The operation of the disk array device in the case of sequential occurrence is shown.

In the figure, 47 to 50 correspond to the read operation of the magnetic disk devices 43 to 46, respectively.

When the host system sequentially issues 47 to 50 read requests, the disk array device internally queues these read requests, and from the executable ones, the built-in magnetic disk actually storing the data. Issue a read request to the device. At this time, even if another read request is already being executed, this request can be executed if the magnetic disk device that actually stores the data requested by the read request is not in operation, and at the same time, it is built in at the maximum. It is possible to execute requests as many as the number of magnetic disk devices. In this way, the data read from the built-in magnetic disk is transferred to the host system in an order irrelevant to the order requested by the host system. For example, in the illustrated example, the read request from the host system is made in the order of d0 to d3, but the data transfer to the host system is done by d2, d1,
This is done in the order of d0 and d3.

As a result, the host system can continue the processing without waiting for the processing of the magnetic disk devices 43 to 46, thus improving the throughput.

FIG. 5 is an explanatory diagram of read access in the case where multi-access is impossible. In this case, a read request from the host system is sequentially issued to all magnetic disk devices in the disk array device. The operation of the disk array device in this case is shown. 51-54 in the figure
Respectively correspond to the read operation of the magnetic disk devices 43 to 46.

In the case of a host system incapable of multi-access, a read request 51 is generated, and after receiving a response to this read request, the next read request 52 is generated. Similarly, read requests 53 and 54 are sequentially generated. In this case, normally, when one magnetic disk device is performing a read operation, the other three magnetic disk devices are in a request waiting state, and efficiency higher than the performance of one magnetic disk device is achieved. Cannot exert.

Further, as shown in FIG. 2, a cache memory 41 is usually mounted on the disk array device in order to speed up read access. The access request from the host system is often relatively sequential access. In this case, the area ahead of the requested area is read into the cache memory 41 in advance, and prefetching is performed to speed up the response to the access. It is possible.

However, in the case of a disk array device, it is desirable that one access be completed by only accessing one built-in disk in order to improve the throughput with respect to multiple accesses from the host system. That is, in the disk array device, when prefetching is performed, it is necessary to suppress the size of a block, which is a unit of distributed arrangement of data to the internal magnetic disk devices 43 to 46. To that end, for a certain read request, the prefetch upper limit may be up to the end of the data block including the requested area. This allows
The performance at the time of sequential access can be improved, and the multi-access from the host system can operate efficiently.

[0015]

As described above, when the prefetch unit to the cache memory 41 is the block unit of each magnetic disk device 43 to 46 in the disk array device, the multi-access request is possible. Valid for host systems. However, for host systems that do not allow multi-access,
The result is that the responses of the magnetic disk devices 43 to 46 in the disk array device are sequentially waited for, so that improvement in performance cannot be expected.
On the other hand, when the prefetch to the cache memory 41 is performed to the area of another magnetic disk device in the disk array device, the performance improvement can be expected for the host system in which the multi-access request is impossible. However, when the host system is multi-accessible, even though the access to one magnetic disk device in the disk array and the prefetch are completed, 2
There is a problem in that an access request is generated to more than one magnetic disk device, and thus the degree of parallel access, which is the original purpose of multi-access, is reduced, leading to a decrease in throughput. It was

From this point of view, it has been desired to realize a disk array device capable of operating efficiently regardless of whether the host system is multiaccessible or not.

[0017]

The present invention employs the following structure to solve the above-mentioned problems. <Structure of claim 1> A plurality of magnetic disk devices are built-in,
Further, in a disk array device having a cache memory, when the request from the host system is multi-access, the access unit to each magnetic disk device is a unit of distributed arrangement of data to each magnetic disk device. If a read request for data stored in one of the magnetic disk devices is made into a block and the block containing that data is stored in the cache memory, and the request from the host system is not multi-access,
When a read request is made for data stored in any of the magnetic disk devices, a block including the data and a cache memory control unit for storing a block continuous to this block in the cache memory are provided. It is characterized by.

<Explanation of Claim 1> According to the invention of Claim 1, when a read request is issued from the host system to the disk array device, the read request is a request by multi-access or a request by other than that. The access method is changed depending on whether or not.

That is, if the request from the host system is multi-access, the cache memory control unit 1
Read in block units and store in cache memory. This makes it possible to take full advantage of multi-access, which is parallel access.

If the read request from the host system is not multi-access, only one block containing the read-requested data is returned to the host system.
In the cache memory, one block containing the requested data and blocks having consecutive addresses are stored simultaneously in this block. Therefore, when an access request is issued from the host system to the data of the next block having consecutive addresses, the data in the cache memory can be used as a response.

As described above, even when the host system is capable of multi-access or incapable of multi-access, efficient operation can be performed as a disk array device.

<Structure of Claim 2> In the disk array device according to Claim 1, when the cache memory control unit includes a tag for identifying which access is included in the access request from the host system. A disk array device characterized by being judged to be multi-access.

<Explanation of Claim 2> When the host system performs multi-access, the tag is transmitted in the access procedure. The tag indicates the order of access, and is required in multi-access in which the order of issuing requests and the order of corresponding responses do not necessarily match. When the cache memory control unit receives an access request from the host system, the cache memory control unit determines whether or not this tag is transmitted, and when the tag is included, determines that the tag is multi-access. This makes it possible to easily determine whether the request from the host system is multi-access.

[0024]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the drawings. <Structure> FIG. 1 is a structural diagram showing a specific example of the disk array device of the present invention. In the figure, reference numeral 100 denotes a disk array control unit, which includes a cache memory 1 and a cache memory control unit 2. Further, 3 and 4 are magnetic disk devices built in the disk array device, and a plurality of magnetic disk devices are provided, but other configurations are omitted in the drawing. Further, 5 and 6 are disk interfaces of the magnetic disk devices 3 and 4, respectively, and 7 is a host interface with the host system.

Here, the cache memory control unit 2 is composed of a control processor, and when the request from the host system is multi-access, the access unit to each magnetic disk device 3, 4 is set to each magnetic disk device 3. ,
One block, which is a unit of data distributed arrangement to four, and a read request for data stored in one of the magnetic disk devices 3 (or 4), a block including the data is cache memory. If the request from the host system is not multi-access,
When a read request is made for data stored in any of the magnetic disk devices 3 (or 4), a function of storing a block including the data and a block continuous to this block in the cache memory 1 is provided. Have Further, the cache memory control unit 2 has a function of determining multi-access when a tag for identifying which access is included in the access request from the host system.

Further, in the figure, 8 and 9 are blocks of data to be read requested by the host system, which are stored in the magnetic disk devices 3 and 4, and the data size of each is stored in each built-in magnetic disk device. One block is a unit of distributed arrangement of data.

<Operation> First, the case where the host system is multi-accessible will be described. When the host system issues a read request for the data in the data block 8 to the disk array device, the cache memory control unit 2 in the disk array device reads the data block 8 from the magnetic disk device 3, Upon completion of this read processing, the data block 8 is stored in the cache memory 1 at the same time as the requested data (data in the data block 8) is responded to the host system. In the figure, a data block 8a in the cache memory 1
Shows a data block in which is stored. Also, when the host system issues a read request for the data in the data block 9, the same process is performed and the data block 9a is stored in the cache memory 1.

Since one read is performed in block units and one access is completed by accessing one built-in magnetic disk device 3 (or 4), it is possible to operate efficiently against multiple access. You can Further, when the requests from the host system are sequentially generated, the requests are concentrated on the data in the cache memory 1 after being stored in the cache memory 1 in units of blocks, so that the disk array device can operate efficiently.

Next, a case where the host system cannot perform multi-access will be described. Data block 8 is sent to this disk array device by a request other than multi-access.
When a read request for the data in the data is issued, the cache memory control unit 2 reads the data blocks 8 and 9 from the magnetic disk devices 3 and 4, respectively, and stores both the data blocks 8 and 9 in the cache memory 1. , The host system responds with the requested data (data in the data block 8). In this case, access to the magnetic disk devices 3 and 4 is executed by multi-access.

When viewed from the host system, the data blocks 8 and 9 are areas where the addresses are continuous. When the access request is sequentially issued from the host system, the data blocks 8 and 9 continue to the next address. Access requests are often made to blocks. Therefore, even if the request from the host system is not multi-access, it can be expected that the host system can efficiently respond.

Next, the operation of the cache memory control unit 2 for determining whether or not the access request from the host system is multi-access will be described.

When the host system performs multi-access in a configuration in which the host system and the disk array device are connected by, for example, a SCSI interface,
The tag is transmitted in the access procedure. Tags are for indicating the order of access,
This is necessary in multi-access in which the order of issuing requests and the order of corresponding responses do not necessarily match. When receiving the access request from the host system, the disk array device determines whether or not this tag is transmitted. This makes it possible to determine whether or not the request from the host system is multi-access.

<Effect> As described above, in the above specific example,
When the request from the host system is multi-access, the data of one block unit is read from one magnetic disk device 3 (or 4) which stores the data,
By storing it in the cache memory 1, it can be expected to respond to a subsequent read request from the host system at high speed. If the request from the host system is not multi-access, the continuous data block is read from the plurality of magnetic disk devices 3 and 4 by multi-access inside the disk array device and stored in the cache memory 1. It can be expected to respond to the host system at high speed. As a result, there is an effect that the access time to the host system can be shortened as compared with the conventional case.

Further, since it is determined whether or not the access request from the host system is multi-access by determining the presence / absence of a tag to be transmitted, whether or not the request from the host system is multi-access. Can be easily discriminated.

[Brief description of drawings]

FIG. 1 is a configuration diagram showing a specific example of a disk array device of the present invention.

FIG. 2 is a diagram showing a typical configuration of a disk array device.

FIG. 3 is an explanatory diagram of an arrangement example of data blocks in the disk array device.

FIG. 4 is an explanatory diagram of read access when multi-access is possible.

FIG. 5 is an explanatory diagram of read access when multiple access is impossible.

[Explanation of symbols]

 1 cache memory 2 cache memory controller 3, 4 magnetic disk device 8, 8a, 9, 9a data block

Claims (2)

[Claims] 1. In a disk array device including a plurality of magnetic disk devices and having a cache memory, when a request from a host system is multi-access, the access unit to each of the magnetic disk devices is One block, which is a unit of distributed arrangement of data to each magnetic disk device, is stored in the cache memory in response to a read request for data stored in one of the magnetic disk devices.
If the request from the host system is not multi-access, and if there is a read request for data stored in any of the magnetic disk devices, a block containing the data and a block continuous to the block are written. A disk array device comprising a cache memory control unit for storing in a cache memory. 2. The disk array device according to claim 1, wherein the cache memory control unit is multi-access when a tag for identifying which access is included in the access request from the host system is included. A disk array device which is characterized by: