JP2006065743A - Disk array device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a storage device capable of preventing a capacity of a cache memory from being used wastefully, while shortening a response time when a disk trouble occurs. <P>SOLUTION: In this disk array device provided with a plurality of the storage devices for storing user data and two or more kinds of redundant data generated from the user data, a control part for controlling an output of the data to the storage device, and a cache for storing temporarily the data inputted and outputted to/from the storage device, the control part reads out the redundant data from the storage device when reading out the user data from the storage device, and stores one kind of redundant data in the cache together with the user data. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a disk array apparatus having a RAID configuration, and more particularly to a stage technique for caching parity data.

  There is known a RAID array disk array device that stores redundant information related to user data stored in a disk array device in a part of the disk array device and reconstructs user data when a failure occurs in the disk. . Among the RAID configurations, in RAID 6, data is divided and stored on a plurality of disks, and two types of redundant data are generated from the data. Each redundant data is stored in a different disk.

Further, the disk array device has a cache memory for temporarily storing data read from the disk (or written to the disk), and shortens the time required for data input / output. At this time, a technique has been proposed in which redundant data is simultaneously read when data is read from the disk, thereby shortening the response time when the data read fails (see, for example, Patent Document 1).
Japanese Patent Application Laid-Open No. 11-212728

  In the prior art described above, in a RAID 5 configuration disk array device, redundant data as well as user data is read from the disk and stored in the cache. When this technology is applied to a RAID 6 configuration disk array device, two types of redundant data are stored in the cache memory. However, in a RAID 6 configuration disk array device, no method has been proposed for restoring data using redundant data stored in a cache when a disk failure occurs.

  In addition, in a RAID 6 configuration disk array device, when two types of generated redundant data are stored in a cache memory, a large-capacity cache memory is required. Since it is rare that a failure occurs in two disks at the same time, if the two types of generated redundant data are always stored in the cache memory, the capacity of the cache memory is wasted.

  The present invention provides a method of restoring data using redundant data stored in a cache when a disk failure occurs in a disk array device having a RAID 6 configuration. Further, the present invention provides a storage device that reduces the response time in the event of a disk failure and prevents waste of the cache memory capacity by changing the redundant data stored in the cache according to the need for redundant data. For the purpose of provision.

  The present invention includes a plurality of storage devices that store user data and two or more types of redundant data generated from the user data, a control unit that manages input / output of data to / from the storage device, and an input to the storage device. And a cache that temporarily stores data to be output. When the controller reads the user data from the storage device, the controller reads the redundant data from the storage device, and the user data At the same time, the one type of redundant data is stored in the cache.

  According to the present invention, in a disk array device having a RAID 6 configuration, it is possible to prevent waste of the capacity of the cache memory while reducing the response time when a disk failure occurs.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram of a computer system according to the first embodiment of this invention.

  This computer system includes a disk array device 100, a plurality of higher-level devices (host computers) 200, and a SAN 300 that connects each device in the computer system.

  The disk array device 100 is provided with a CPU 110, a memory 120, a cache 130, an upper interface 140, a disk interface 150, and a disk 160.

  The CPU 110 performs various processes by loading and executing the management programs 121 to 124 stored in the memory 120. For example, data transfer among the upper interface 140, the disk interface 150, and the cache memory 130 is controlled.

  The memory 120 is a volatile or non-volatile memory, and stores a cache management program 121, a disk input / output management program 122, a RAID management program 123, and a parity generation management program 124. In addition, a work area necessary for calculation by the CPU 110 is provided.

  The cache 130 is a memory that temporarily stores data input and output from the disk array device 100. The cache 130 is provided with a data area 131 and a parity area 132.

  The data area 131 stores data transferred between the upper interface 140 and the disk interface 150. In the parity area 132, redundant data (parity) corresponding to the data stored in the data area 131 is stored. The cache 130 stores a cache management table 133 (FIG. 2).

  Note that the cache 130 may be provided in a storage element separate from the memory 120, but a partial area of the same storage element as the memory 120 may be used as the cache 130.

  The host interface 140 is an interface to the host device 200 and transmits / receives data and control signals to / from the host device 200 via the SAN 300. The communication with the host device 200 uses a fiber channel protocol.

  The disk interface 150 is an interface to the disk 160, and transmits and receives data and control signals to and from the disk by a protocol such as ATA, SAS, or fiber channel.

  The disk 160 is a storage device that stores data, and includes, for example, a plurality of magnetic disk drives. The disk 160 constitutes a RAID (Redundant Array of Independent Disks), and the user data is divided into a plurality of disks and input / output to improve the read / write performance of the disk. Further, the stored data is provided with redundancy so that the stored data is not lost even if a failure occurs in a part of the disk.

  The disk 160 is divided into logical partitions and constitutes a logical unit. Alternatively, a logical unit may be configured from a plurality of disk devices 160.

  Furthermore, in this embodiment, the disk 160 constitutes RAID6. That is, user data is divided and stored on a plurality of disks 160, and each of the two types of redundant data (parity) generated from the user data is stored in different disks 160 constituting a parity group.

  The host device 200 is a computer device having a CPU, a memory, a storage device, and an interface. The host device 200 provides a database service, a web service, and the like using data provided from the disk array device 100.

  The SAN 300 connects the disk array device 100 and the host device 200. The SAN 300 is a network that can communicate with a protocol suitable for data transfer, such as a fiber channel protocol.

  Next, various management programs stored in the memory 120 of the disk array device 100 will be described.

  The cache management program 121 refers to the cache management table 133 and manages the area of the cache 130 in which data is stored. It also manages in which area of the cache 130 the data is stored. It also manages which data the redundant data stored in the parity area corresponds to.

  The disk input / output management program 122 manages data transfer between the upper interface 140, the disk interface 150, and the cache 130. Further, data is read from and written to the disk 160 in accordance with a data input / output request from the host device 200.

  The RAID management program 123 sets a RAID on the disk 160. In addition, a parity group is configured by a plurality of disks 160 to control RAID operations. That is, the data is distributed and stored in a plurality of disks in the same parity group, and the data distributed and stored in the plurality of disks is read and reconstructed. Even if one disk fails, lost data is restored by an error correction function (collection).

  The parity generation program 124 generates redundant data (parity) based on the data stored in the disk. In the present embodiment, the disk array device 100 has a RAID 6 configuration. Therefore, the parity generation program 124 includes two parity generation programs A and B. Two types of parity are generated by the two parity generation programs.

  The above-described programs 121 to 124 are executed by the CPU 110, but the functions of these programs may be configured by hardware.

  FIG. 2 is a configuration diagram of the cache management table 133 according to the first embodiment.

  The cache management table 133 stores a cache address, a disk address, and a data status flag.

  The cache address is an address on the cache memory, and stores a value indicating the relationship between data stored in the cache and parity (for example, a pair of a data address and a parity address). Depending on which area the cache address is contained in, it can be determined whether it is data or parity.

  The disk address is an address on the disk corresponding to the data stored in the cache.

  The data state flag is information indicating a generation state of data stored in the cache. For example, if the data status flag is “0”, the data stored in the cache is the staging data read from the disk, and if the data status flag is “1”, the data is stored in the cache. Indicates that the stored data is restored on the cache.

  The restored data can be specified by the data status flag, and destage processing for writing the restored data to the disk 160 when the disk is recovered from a failure can be performed.

  FIG. 3 is a conceptual diagram of data read processing in the disk array device 100 according to the first embodiment.

  Data (D 1, D 2) related to the read request read from the disk 160 is temporarily stored in the data area 131 of the cache 130 and transferred to the host device 200.

  Simultaneously with the reading of this data, the two parities (P, Q) of the data relating to the read request are read from the disk and stored in the parity area 132 of the cache 130. Note that the parity read from the disk 160 is not transferred to the higher-level device 200.

  FIG. 4 is a flowchart of data writing processing according to the first embodiment.

  First, when receiving a data write request from the host device 200 (S111), the disk input / output management program 122 stores data (new data) related to the write request in the data area 131 of the cache 130 (S112).

  When the new data is stored in the cache 130, the disk input / output management program 122 transmits a data write completion report to the higher-level device 200 (S113). Writing new data to the disk 160 is performed asynchronously by a write-after process after a data write completion report to the host device 200 is made.

  After that, the parity generation management program 124 identifies the address of the disk 160 where the data written from the host device 200 is stored, and reads the old data and old parity written at the address from the disk 160 (S114).

  Then, the parity generation management program 124 generates a new parity using the old data read from the disk 160, the old parity, and the new data written from the higher-level device 200 (S115).

  Thereafter, the disk input / output management program 122 reads the new data from the cache 130 and writes it to the disk 160 together with the generated new parity (S116).

  FIG. 5 is a conceptual diagram of the data writing process (FIG. 4) according to the first embodiment.

  New data (ND1, ND2) written from the host device 200 is stored in the data area 131. Thereafter, old data (D1, D2) and old parity (P, Q) corresponding to the new data are read from the disk 160.

  The parity generation management programs A and B generate new parity (NP, NQ) using the old data read from the disk 160, the old parity, and the new data written from the host device 200.

  At this time, the parity generation program A calculates the parity P by exclusive OR of the data stored in the disk 160. In addition, the parity generation program B calculates the parity Q by Galois calculation of data stored in the disk 160. Note that the Galois operation has a larger calculation amount for parity generation and data restoration using parity than the exclusive OR operation. Therefore, when only one type of parity is used as will be described later, the parity P calculated by exclusive OR may be used.

  FIG. 6 is a flowchart of processing when a failure occurs in the parity on the cache according to the first embodiment.

  First, when the disk input / output management program 122 receives a data read request from the host device 200 (S121), the disk in which the data related to the request is stored is specified from the data read request (S122). Then, with reference to a table managed in the disk array device 100, it is determined whether or not a failure has occurred in the identified disk (S123).

  As a result, if no failure has occurred in the disk, normal read processing is performed. That is, the data related to the data read request and one parity P of the data are read from the disk 160. Thereafter, the read data and one parity P of the data are stored in the cache 130 (S124).

  One parity to be read at this time may be any parity, but it is preferable to read a parity that can easily restore data (for example, a parity P by an XOR operation).

  On the other hand, if a failure has occurred in the disk, all readable redundant data (parity) is read out. Before that, for each data area 131 and parity area 132, the table managed in the disk array device 100 is referred to determine which data is stored in the disk, and which data has failed. Alternatively, it is determined which parity can be read (S125).

  As a result, if a failure has occurred in the disk in which the data is stored, the data of the disk in which no failure has occurred and the two parities P and Q of the data are read from the disk 160. Thereafter, the read data and the two parities P and Q of the data are stored in the cache 130 (S126).

  Then, the RAID management program 123 restores the data on the cache using the data and the parity P. Then, the restored data is stored in the cache 130 and transferred to the host device 200 (S127). Thereafter, the process proceeds to step S131. Note that the parities P and Q read from the disk 160 are not transferred to the higher-level device 200.

  On the other hand, if a failure has occurred in the disk in which the parity P is stored, the data related to the data read request and one parity Q of the data are read from the disk 160. Thereafter, the read data and one parity Q of the data are stored in the cache 130 (S128).

  According to the above processing, at the normal time when no failure occurs in the disk 160, two parities are generated by RAID 6 and written to the disk 160, but only one parity is read as in RAID 5 (S124). However, in normal times when no failure has occurred in the disk 160, two parities are generated by RAID6 and written to the disk 160, but only one parity is read as in RAID5 (S126, S128). In this way, by controlling writing and reading in different modes, it is possible to suppress a decrease in normal reading performance.

  In step S129, it is determined whether reading of data from the cache 130 has failed.

  As a result, if the reading of data from the cache 130 has not failed, this process ends.

  On the other hand, if the reading of data from the cache 130 has failed, it is determined whether reading of either data or one parity has failed (S130).

  On the other hand, if the reading of data or one parity has failed, it is further determined whether reading of data or reading of one parity has failed (S131).

  As a result, if the data reading has failed, the data is restored on the cache using the data and parity stored in the cache (S132), and the process proceeds to step S134.

  On the other hand, if reading of one parity fails, the parity is recalculated on the cache using the data stored in the cache. The recalculated parity is stored in the cache 130 (S133), and the process proceeds to step S134.

  Then, the data related to the read request is transferred to the host device 200 (S134).

  Next, a modification of the first embodiment will be described.

  In this modification, the configuration of the cache 130 is different.

  In the above-described embodiment (FIG. 1), the cache 130 is divided by the address, and the data area 131 and the parity area 132 are provided. However, in the second modification, the data area 131 and the parity area 132 are not provided in the cache 130. For this reason, the information stored in the cache management table 133 is different.

  FIG. 7 is a configuration diagram of the cache management table 133 according to a modification of the first embodiment.

  The cache management table 133 stores a cache address, a disk address, a data status flag, and a data / parity flag.

  The cache address, disk address, and data status flag are the same as those in the cache management table (FIG. 2) described above.

  The data / parity flag is information indicating whether the content stored in the location indicated by the cache address is data or parity. For example, if the data / parity flag is “0”, the data is stored, and if “1”, the parity is stored.

  The data / parity flag can specify whether the content stored in the cache is data or parity. Therefore, it is not necessary to provide a data area and a parity area in the cache 130, and the cache 130 can be used efficiently.

  In the embodiment of the present invention, it is possible to shorten the response time when a disk failure occurs in a disk array device having a RAID 6 configuration. Therefore, even in a disk array device that is used for broadcasting and reading video and audio streaming data, it is possible to prevent video corruption and audio interruption due to a disk failure.

  Further, since the parity stored in the cache is changed according to the state of the disk failure, it is possible to prevent waste of the capacity of the cache memory. Therefore, it is possible to provide a disk array device that has a small cache capacity, is highly reliable, and is inexpensive.

1 is a block diagram of a computer system according to a first embodiment of this invention. 3 is a configuration diagram of a cache management table 133 according to the first embodiment. FIG. It is a conceptual diagram of the data read-out process of 1st Embodiment. It is a flowchart of the data writing process of embodiment of 1st invention. It is a conceptual diagram of the data writing process of embodiment of 1st this invention. 6 is a flowchart of data read processing when a failure occurs in parity on a cache according to the first embodiment; It is a block diagram of the cache management table 133 of the modification of 1st Embodiment.

Explanation of symbols

100 disk array device 110 CPU
120 Memory 130 Cache 133 Cache Management Table 133
140 Host interface 150 Drive interface 160 Disk 200 Host device 200
300 SAN

Claims (7)

A plurality of storage devices that store user data and two or more types of redundant data generated from the user data, a control unit that manages input / output of data to / from the storage device, and data input / output to / from the storage device A disk array device comprising a cache for temporarily storing
The controller is
When reading the user data from the storage device, read the redundant data from the storage device,
A disk array device that stores the one type of redundant data in the cache together with the user data.   The disk array device according to claim 1, wherein the control unit reads the one type of redundant data from the storage device when reading the user data from the storage device. The controller is
Normally, the one type of redundant data is stored in the cache,
2. The disk array device according to claim 1, wherein when the failure occurs in the storage device, the two types of redundant data are stored in the cache. A plurality of storage devices that store user data and two or more types of redundant data generated from the user data, a control unit that manages input / output of data to / from the storage device, and data input / output to / from the storage device A disk array device comprising a cache for temporarily storing
The controller is
Storing the one type of redundant data stored in the storage device in the cache;
A disk array device that restores user data with redundant data read from the cache when the user data is restored. The controller is
Normally, the one type of redundant data is stored in the cache,
When a failure occurs in the storage device, at the time of restoration of the user data, the one type of redundant data is read from the cache, and the user data is restored by the read one type of redundant data. The disk array device according to claim 4, wherein the redundant data is stored in the cache. A plurality of storage devices that store user data and two or more types of redundant data generated from the user data, a control unit that manages input / output of data to / from the storage device, and data input / output to / from the storage device A disk array device comprising a cache for temporarily storing
The controller is
When reading user data from the storage device, the two types of redundant data are read from the storage device,
A disk array device for storing the two types of redundant data in the cache together with the user data. The controller is
When a failure occurs in the storage device, the one type of redundant data is read from the cache when the user data is restored,
The disk array device according to claim 6, wherein user data is restored by the read one type of redundant data.

Images