JP2001100937A - Disk array controller - Google Patents

Abstract

PROBLEM TO BE SOLVED: To make the read-in speed of data fast when new data are written and to efficiently generate parity data. SOLUTION: This controller is provided with a judging means 7 that selects a means for reading less disk drives 5 between a 1st and a 2nd means by comparing the number of disk drives 5 which need to read in data by the 1st means generating new parity data from newly rewritten data and data of other identical stripes when the data of the same stripes are rewritten with the number of disk drives 5 which need to read in data by the 2nd means generating new parity data from the old data before the rewriting and old parity data and performing weighting processing so that the 1st means is easy to select.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a disk array controller, and more particularly, to a disk array controller for performing data access to a plurality of disk devices in parallel.

[0002]

2. Description of the Related Art Conventionally, a large amount of data can be managed by connecting a plurality of disk devices such as hard disk drives via a parallel interface, and at the same time, data access is speeded up and data is stabilized. RAID level disk array devices have been developed.

For example, in such a disk array device of RAID level 5, data is distributed and recorded in a stripe form on each disk device, and parity data of each data is also distributed and recorded on each disk device. This enables high-speed access by performing data access in parallel,
It is also possible to recover data based on the parity data.

In such a conventional disk array device, when data in the same stripe of each disk device is rewritten, parity data is newly created. As means for creating new parity data, there are means (first means) for creating new parity data from newly rewritten data and other data in the same stripe, and means for newly rewriting data. There is a means (second means) for creating new parity data from the updated data, the old data before rewriting, and the old parity data.

In general, when data of each disk device is read, this data is recorded in a cache memory, and the subsequent data is read not from the disk device but from the cache memory so as to increase the reading speed. Has become.

Therefore, whether to create new parity data by the first means or to create new parity data by the second means is determined as follows.
Conventionally, data recorded in the cache memory is first examined, and the number of disk devices that need to read data not recorded in the cache memory is compared with each other. In other words, as shown in FIG. 4, the number n1 of disk devices for which data must be newly read into the parity by the first means, and the data is read when new parity data is created by the second means. Compare the number n2 of required disk units with
Select the means with the smaller number of disk devices to read,
New parity data was created.

[0007]

However, in the conventional parity data generating means, the parity data is generated by selecting means having a small number of disk devices which need to be read. When the data is recorded, the data in the cache memory can be used even when the data is subsequently rewritten, so that the data necessary for creating the parity data is recorded in the cache memory. The parity data can be created more efficiently by using the first means that is more likely to be performed.

However, the conventional parity data generating means merely compares the number of disk devices from which data is read, so that the data reading speed cannot always be increased, and the parity data is generated efficiently. Have the problem of not being able to do so.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and when a new data is written, a disk array control device capable of increasing the data reading speed and efficiently generating parity data. The purpose is to provide.

[0010]

In order to achieve the above object, a disk array control apparatus according to the first aspect of the present invention connects a plurality of disk devices via a parallel interface, and sets the same disk device as the same. When the stripe data is rewritten, the newly rewritten data,
A first means for creating new parity data from other data of the same stripe, and a second means for creating new parity data from newly rewritten data, old data before rewriting, and old parity data A disk array control device in which any of the means is selected to newly generate parity data, wherein the number of disk devices for which data must be read by the first means; The second means is compared with the number of disk devices that need to read data, and the first means is weighted so as to be more easily selected. It is characterized in that a judging means for selecting a means is provided.

According to the first aspect of the present invention, the number of disk devices that need to read data by the first means and the number of disk devices that need to read data by the second means are determined by the determination means. Compare with the number of devices,
Weighting is performed so that the first means is easier to select, and the means with the smaller number of disk devices to be read is selected, so that the chance of creating parity data by the first means is increased. In such a case, the data in the cache memory can be used even when the data is subsequently rewritten.
The data reading speed can be increased, and parity data can be created quickly.

Further, in the disk array control device according to the present invention, the weighting is based on the number of disk devices that need to read data by the first means. This is characterized in that the processing is performed by reducing an arbitrary number corresponding to the number.

According to the second aspect of the present invention, the first
The weighting is performed by subtracting an arbitrary number corresponding to the number of disk devices from the number of disk devices that need to read data by means of (1), so that the first means can be selected appropriately. It can be controlled to be easy.

[0014]

FIG. 1 is a block diagram showing an embodiment of the present invention.
3 will be described with reference to FIG.

FIG. 1 shows an embodiment of a disk array control device according to the present invention. The disk array control device 1 of the present embodiment outputs a command such as data read or write to It has a host device 2 for making an O request, and this host device 2 has
The RAID controller 4 is connected via the host bus 3.

A plurality of disk devices 5, 5..., Such as hard disk drives, for example, of the SCSI interface type, are connected to the RAID control device 4 in parallel via a drive bus 6, respectively.

Further, the RAID control device 4 newly generates parity data when data of the same stripe is rewritten. In this case, for example, as shown in FIG. 2, when D0, D1, D2, and D3 are recorded as stripe data of each disk device 5, one stripe data D0 is newly rewritten to D0 '. In such a case, there are the following means for creating new parity data P 'from the parity data P.

First, of the same stripe, newly rewritten data D0 'and other data D1, D2,.
There is a means (first means) for creating new parity data P 'from D3 based on the following equation (1). (1) P ′ = D0 ′ + D1 + D2 + D3 In addition, newly rewritten data D0 ′,
There is a means (second means) for creating new parity data P ′ from the old data D0 before rewriting and the old parity data P based on the following equation (2). (2) P ′ = D0 ′ + D0 + P Further, in the present embodiment, the RAID controller 4
The determination means 7 is provided to generate new parity data P 'by the first means based on the above equation (1), or the second means based on the equation (2). It is determined whether or not new parity data P 'is to be created by the means. More specifically, in the present embodiment, when making this determination, the number n1 of the disk devices 5 that need to read the data when the parity data is newly created by the first means, and the second When the parity data is newly created by the means, the number n2 of the disk devices 5 from which the data needs to be read is compared with the number n2 of the disk devices 5, and the means of the smaller number of the disk devices 5 to be read is selected. In this case, in the present embodiment, the first means is more easily selected.
Weighting is performed.

In the case of the first means, since the value required for the expression (1) uses data of the same stripe, there is a high possibility that each data is recorded in the cache memory. In some cases, the reading of the disk device 5 is unnecessary, whereas in the case of the second means, the old data and the old parity data are used, and the old data and the old parity data are stored in the cache memory. Since no data is recorded, in the case of the second means, it is necessary to read the disk device 5 without fail, so that the first means is more likely to generate parity data more quickly.

Next, the operation of the present embodiment will be described with reference to the flowchart shown in FIG.

In this embodiment, when new stripe data is written to the disk device 5, the judging means 7 creates new parity data by either the first means or the second means. Is determined. In this embodiment, the determination is made by the determination means 7 by the value obtained by subtracting the weight g from the number n1 of the disk devices 5 which need to read data in the case of the first means, and by the second means Is compared with the number n2 of the disk devices 5 which need to read the data, and the smaller means is selected. The weight g is appropriately set depending on whether or not a continuous write request is made, the number of disk devices 5, the size of stripe data, and the like. For example, the weight g is set as g = 1.

When the weight g is set to 1, the number n1 of the read disk devices 5 by the first means is
However, even when the number of read disk devices 5 is one more than the number n2 of read disks by the second means, parity data is created by the first means.

Therefore, in the present embodiment, the number of disk drives 5 for reading data by the first means is weighted by the judging means 7 so that the chance of creating parity data by the first means is increased. Therefore, even when data is subsequently rewritten, the data in the cache memory can be used, the data reading speed can be increased, and the parity data can be created quickly. Can do it.

It should be noted that the present invention is not limited to the above-described embodiment, but can be variously modified as needed.

[0025]

As described above, the disk array controller according to the first aspect of the present invention performs weighting so that the first means is more easily selected.
Means to increase the chances of creating parity data by means of (1), so that even when data is subsequently rewritten, data in the cache memory can be used, and the data reading speed can be reduced. Therefore, parity data can be created quickly.

Further, the disk array control device according to the second aspect of the present invention can reduce the weight by subtracting an arbitrary number corresponding to the number of disk devices from the number of disk devices for which data must be read by the first means. Since the attachment is performed, the first means can be controlled so as to be appropriately selected more easily.

[Brief description of the drawings]

FIG. 1 is a block diagram showing an embodiment of a disk array control device according to the present invention.

FIG. 2 is an explanatory diagram showing a data write state by the disk array control device of the present invention.

FIG. 3 is a flowchart showing an operation of creating parity data by the disk array control device of the present invention.

FIG. 4 is a flowchart showing a conventional parity data creation operation;

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 disk array controller 2 host device 3 host bus 4 RAID controller 5 disk device 6 drive bus 7 determination means

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Kenichi Murakami, Inventor Kenichi Murakami 1-7, Yukitani Otsukacho, Ota-ku, Tokyo Alps Electric Co., Ltd. (72) Inventor Hiroyoshi Usui 1-7, Yukitani-Otsukacho, Ota-ku, Tokyo Alp (72) Inventor Mitsuru Maruyama 1-7 Yukitani-Otsuka-cho, Ota-ku, Tokyo Alps Electric Co., Ltd. (72) Inventor Yuichiro Sawame 1-7 Yukitani-Otsuka-cho, Ota-ku, Tokyo Alps Electric Inside (72) Inventor Shigeru Araki 1-7 Yukitani Otsuka-cho, Ota-ku, Tokyo Alps Electric Co., Ltd. (72) Inventor Kentaro Inoue 1-7 Yukitani-Otsuka-cho, Ota-ku, Tokyo Alps Electric Co., Ltd. (72) Inventor Takechiyo Takatsuki 1-7 Yukitani Otsuka-cho, Ota-ku, Tokyo Alps Electric Co., Ltd. F-term (reference) 5B065 BA01 CA07 CA3 0 CC08 EA01

Claims (2)

[Claims] When a plurality of disk devices are connected via a parallel interface and data of the same stripe of each of the disk devices is rewritten, newly rewritten data and other data of the same stripe are newly rewritten. Any one of a first means for creating new parity data and a second means for creating new parity data from newly rewritten data, old data before the rewriting, and old parity data. Is selected to create new parity data, wherein the number of disk devices for which data needs to be read by the first means and the data read by the second means. In addition to comparing the number of disk devices that need to be loaded, weighting is performed so that the first means is more easily selected, and the weight is read. A disk array control device comprising: a determination unit that selects a unit having a smaller number of disk devices. 2. The method according to claim 1, wherein the weighting is performed by subtracting an arbitrary number corresponding to the number of disk devices from the number of disk devices from which data must be read by the first means. 2. The disk array control device according to claim 1, wherein: