JP2006268286A - Disk array device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a disk array device capable of suppressing increase of time used for data restoration and increase of the number of physical disks, providing large capacity, improving reliability, increasing recording speed, reducing costs, and reducing size. <P>SOLUTION: Each of physical disks 11-16 is divided into three areas which are an A block 11A-16B, a B block 11B-16B, and a C block 11C-16C with storage areas that are respectively the same storage capacity. The A blocks 11A-16A are set in relationships comprising sets with the B blocks 11B-16B of different physical disks 11-16 with the same capacity, and the C blocks 11C-16C of physical disks 11-16 that are different to the other two physical disks. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a disk array apparatus, and more particularly to a disk array apparatus that can achieve a large capacity and improved reliability, a high recording speed, a low cost, and a small size.

  In recent years, computers have been introduced not only to companies but also to general households, and the degree of dependence on IT (Information technology) has increased. Accordingly, the importance of electronic data handled by computer systems is increasing, and the role of storage devices in computer systems that store such electronic data is increasing. For this reason, the demand for higher reliability is increasing as well as higher performance and capacity of storage devices.

  In order to maintain the reliability of electronic data as the storage capacity increases, the electronic data stored on the disk should be copied to a spare disk in preparation for failure of the disk with the large storage capacity. Has been performed conventionally. However, in such a method, it is necessary to copy the electronic data saved in the spare disk to the main disk again after repairing the failure of the disk (main disk). Since the time spent for copying is proportional to the storage capacity of the disk, there arises a problem that the time required increases as the storage capacity increases.

  In contrast, in order to improve data read / write speed while maintaining data redundancy, most current hard disks are composed of multiple disks, and data is divided into each disk. A method of storing (so-called RAID: Redundant Arrays of Inexpensive Disks) is employed. A failure countermeasure technique for such multiple hard disks, that is, a technique for ensuring data redundancy has been studied.

  Next, a conventional disk array device will be described with reference to the drawings.

  FIG. 5 is a schematic block diagram showing the technique described in Patent Document 1. As shown in FIG.

  Referring to FIG. 5, the disk array device 200 receives data from the host device 300 and records data by the RAID system under the control of the RAID controller 250. Specifically, FIG. 5 shows a RAID 5 system. In this system, write data is divided into the number of physical disks, and parity data is generated in correspondence with the divided data. The write data and the parity data as a set are recorded on the six physical disks 210 to 215, respectively. However, these six physical disks are provided with separate physical disks 220 to 225 which form pairs. That is, the two units constitute a pair of disks. The pair of disks are two physical disks surrounded by a virtual line in FIG. 5, for example, disks denoted by reference numerals 210 and 220 correspond to these. Then, the write data DA1 to DA6 recorded in the data area of one of the physical disks 210 to 215 is copied to the other physical disks 220 to 225 that form a pair. Therefore, the mirror data DA1 'to DA6' having the same contents as the data DA1 to DA6, which are the divided write data and parity data, are recorded on another physical disk as a pair.

  FIG. 6 is a schematic block diagram showing the technique described in Patent Document 2. As shown in FIG.

  Referring to FIG. 6, this also uses the RAID5 system, and when data is recorded by a command from the host device 500, the RAID controller 450 divides the data into the number of the six physical disks shown. In addition, parity data is generated in correspondence with the divided data, and the divided write data and parity data are recorded as a set on six physical disks DA1 to DA6 (410). The disk array device 400 uses a redundant disk system, and includes a column-direction redundant disk drive 420 and a row-direction redundant disk drive 430 with respect to the data disk drive 410 as shown in the figure. Yes. Then, parity in the row and column directions of the data disk drive 410 in which the divided data and parity data are stored is generated and stored in the column direction redundant disk drive 420 and the row direction redundant disk drive 430. By doing so, when a plurality of physical disks out of the data disk drive 410 fail, the data can be restored more reliably.

JP-A-7-129331 JP 2000-148409 A

  The technique described in Patent Document 1 described above increases the cost of the disk device due to the increase in the number of physical disks, and also increases the space occupied by the physical disk, and the disk array device itself cannot be reduced in size. There is a problem.

  Further, when two physical disks fail, for example, when a failure occurs in a pair (two) of physical disks 210 and 220 in FIG. 5, degeneration occurs, and data recovery processing in a non-redundant state must be performed. There is a problem.

  Since the technology described in Patent Document 2 has a large number of physical disks, as in Patent Document 1, there is a problem of an increase in the cost spent on the disks and a problem that the size cannot be reduced with an increase in storage space. is there.

  When one physical disk fails, the data corresponding to the data stored in the physical disk is temporarily copied to the redundant disk, and then the replacement is completed from the redundant disk to complete the reconstruction. Therefore, there is a problem that a non-redundant state exists in the method using a redundant disk.

  Therefore, a disk array device that can suppress an increase in time spent for data recovery and an increase in the number of physical disks is desired.

  SUMMARY OF THE INVENTION An object of the present invention is to provide a disk array device that solves the above-described problems and can achieve a large capacity and improved reliability, a high recording speed, a low cost, and a small size.

  The disk array device of the present invention has a plurality of physical disks and a control unit that reads / writes data from / to the storage area of the physical disk based on a command from the host apparatus, and each physical disk has a storage area. Each of the three areas having the same storage capacity is divided into a first block, a second block, and a third block. The first block is a second block of another physical disk having the same capacity and two physical disks. Another third block is set in a set relation.

  The set relationship is as follows: one storage area divided from one physical disk, one storage area divided from another physical disk, and one storage area divided from a physical disk different from the other physical disk One storage area of one physical disk and another storage area of another physical disk and one storage area of another physical disk that is different from the other physical disk. Another storage area that is set between the other physical disk and another physical disk that is divided between another physical disk and another physical disk that is divided between the other physical disks. It is characterized by being set between the area.

  The control unit has correspondence data for associating storage areas to be set, and has a management function for holding the same data in the storage areas of the relation to be set based on the correspondence relation data .

  The storage area to be set is set as an area for writing data, and the control unit has write area specifying data for specifying the storage area as the data writing area, and write data according to a command from the host device. Based on the write area specifying data, the same data as the data written in the storage area is copied to another storage area that is a set with the storage area.

  The control unit has a function of reading the write data from any one of the storage areas to be set based on the correspondence data.

  The control unit has a replacement recognition function for recognizing the replacement of the physical disk, recognizes the replacement of the physical disk, and stores the data in the storage area of the other physical disk that is in a set relationship. It has a function of copying to an area.

  In the disk array device of the present invention, when a physical disk fails, the data in the physical disk is copied to another physical disk. Therefore, the physical disk can be replaced while maintaining the redundant state, which is dangerous. Data reliability is increased without a non-redundant period. Therefore, in the conventional disk device, when a large-capacity physical disk is degraded due to a failure, it becomes a non-redundant state until copying to a redundant disk (hot spare, etc.) is completed. In contrast to this, the present invention has a very excellent effect.

  In addition, despite the above-described redundancy and high reliability, the increase in the number of unnecessary physical disks is suppressed, and a low-cost and compact device is realized. it can. Therefore, as the number of physical disks used for data recording increases, the fault tolerance performance against multiple failures can be improved and the cost can be reduced. .

  Next, the best mode for carrying out the present invention will be described in detail with reference to the drawings.

  In FIG. 1, a disk array device 1 is a disk array device 1 that is connected to a host device 30 and reads / writes data based on a command from the host device 30, and is composed of a plurality of physical disks 11-16. A disk drive 10 and a control unit 20 (RAID controller) that controls reading and writing of data with respect to the physical disks 11 to 16 are provided. That is, as will be described later, the control unit 20 has the correspondence data for associating the storage areas to be set, and based on the correspondence data, the same data is stored in the storage areas of the relation to be set. It has a management function to hold.

  In the disk drive 10, six physical disks 11 to 16 are stored in FIG. Each of the physical disks 11 to 16 is set in advance by dividing the data storage area into three areas having the same storage capacity. Here, the divided areas of the physical disks 11 to 16 are called blocks, and the three areas are called A block, B block, and C block, respectively. The divided A block, B block, and C block are set to the same capacity. That is, each of the physical disks 11 to 16 has divided areas 11A and 11B and 11C, 12A and 12B and 12C, 13A and 13B and 13C, 14A and 14B and 14C, and 15A having a data storage area of 1/3. It is divided into 15B and 15C, and 16A, 16B and 16C.

  Further, in the present invention, the divided storage area of one physical disk includes a set of a divided storage area of another physical disk and a divided storage area of another physical disk different from the other physical disk. The relationship is established. An example of the relationship forming the set will be described. The relationship forming the set is as shown in FIGS. 3A and 3B described later, and the divided storage area of the one physical disk and the other Set between a divided storage area of the physical disk and a divided area of a physical disk different from the other physical disk, and another divided storage area of the one physical disk , Set between another storage area of the other physical disk and one storage area of another physical disk that is different from the physical disk different from the other physical disk, and The remaining storage area of the one physical disk is further separated from the one storage area of the physical disk different from any of the physical disks and the physical disk different from any of the physical disks. Physical display It is set between the divided first storage regions.

  In the above example, one divided storage area of the one physical disk, one divided storage area of the other physical disk, and one divided storage of a physical disk different from the other physical disk A description will be given of the relationship that forms a set with the area.

  In this case, any one divided area forming a set in which data of the same content is recorded is set to the A block of one physical disk, and the other two divided areas forming the set are the same as the one physical disk. Are set to B blocks of different physical disks and C blocks of physical disks different from the other physical disks. In addition, the A block of the other physical disk is set to form a set with a B block of a physical disk having the C block as a storage area and a C block of another physical disk different from the other physical disk. Has been.

  Furthermore, it demonstrates concretely with reference to FIG. The physical disk denoted by reference numeral 12 will be described. The divided areas forming a set in which data of the same contents are stored are the A block (12A) of one physical disk 12, the B block (13B) of another physical disk 13, and the above-described divided areas. It is set in the C block (14C) of the physical disk 14 different from the other physical disks. Accordingly, the A block (12A) of the physical disk 12, the B block (13B) of the physical disk 13, and the C block (14C) of the physical disk 14 form a set, and these A block (12A) and B block ( 13B) and C block (14C) are written with the same contents. Also, the A block (13A) of the physical disk 13, the B block (14B) of the other physical disk 14, and the C block (15C) of the physical disk 15 different from the other physical disks have the same data. It is set as a divided area forming a set to be written. Similarly, as shown in FIG. 2, physical blocks 14, 15, 16, and 11 A blocks 14A, 15A, 16A, and 11A and physical disks 15, 16, 11, and 12 B blocks 15B, 16B, 11B, and 12B and physical The C blocks 16C, 11C, 12C, and 13C of the disks 16, 11, 12, and 13 are set as divided areas that are sets in which data having the same contents are written. In addition, the relationship between the A block, B block, and C block to be set is indicated by arrows Y1 to Y6.

  Next, the operation of the best mode for carrying out the present invention will be described with reference to the drawings.

  Correspondence data between the divided areas constituting the set indicating the corresponding relation with the A block, B block, and C block set as the divided areas constituting the set described above is a storage area (non-volatile) in the RAID controller 20 as the control unit. Memory). The RAID controller 20 determines the A block of one physical disk 11 to 16, the B block of other physical disks 12, 13, 14, 15, 16, and 11, and the other physical disk based on the correspondence data. A management function for holding the same data in the C blocks of other physical disks 13, 14, 15, 16, 11, 12 is executed.

  Here, the RAID controller 20 as a control unit includes a RAID control unit 21 that manages RAID construction, a disk driver 22 that manages access to a physical disk, and a mirror control unit 23 that performs copying. Yes. In particular, the mirror control unit 23 stores the divided areas (A, B, and C blocks) of the physical disks 11 to 16 as a set based on the correspondence data stored in the RAID controller 20. On the other hand, a function for writing the same data, that is, a mirroring execution function is provided.

  Of the RAID systems, the RAID 1 and RAID 5 systems will be described in particular. The RAID 1 method is a method for protecting the data by writing the same data to two physical disks and dealing with the occurrence of a failure in one physical disk. The RAID5 method is a method in which write data is divided into a plurality of pieces, parity data is generated corresponding to the divided write data, and is distributed and stored in a plurality of physical disks.

  In the disk array device of the present invention, any one of the divided areas to be set is set as a data writing area. For example, when the divided area 12A of the physical disk 12, the divided area 13B of the physical disk 13, and the divided area 14C of the physical disk 14 are set, one of the divided areas 12A is set as the data writing area. Is done. In such a set of divided areas, write area specifying data indicating that the divided area 12A is a data writing area is recorded in a storage area in the RAID controller 20. The write area specifying data is stored in a memory in the disk driver 22, for example. The same applies to the other divided areas that form a set.

  Accordingly, the RAID controller 20, for example, the disk driver 22 has a function of writing write data to the data write area in accordance with a command from the host device 30 based on the write area specifying data. ing. Based on the correspondence data representing the correspondence of the divided areas to be set by the mirroring execution function described above, the same data as the data in the data writing area is divided into the divided areas to be set with the area. Written. By writing data to one of the physical disks that make a set in this way, the written data is copied to the other two physical disks by the mirroring function, so the data write becomes a set. Only one of the physical disks needs to be performed, and data writing speed can be increased while maintaining data redundancy.

  Next, the operation of the mirroring function will be described in detail with reference to FIG.

  FIG. 2 shows the disk drive disclosed in FIG. DA1, DA2, DA3, etc. are written in each block 11A, 11B, 11C, etc. of each physical disk 11-16 in FIG. For example, the storage area of the physical disk 12 is divided into an A block (12A), a B block (12B), and a C block (12C) having the same capacity as described above, and the A block (12A). Is a set of the B block (13B) of the physical disk 13 and the C block (14C) of the physical disk 14. (See arrows Y2, Y2 '). The B block (12B) of the physical disk 12 is a set of the C block (13C) of the physical disk 13 and the A block (11A) of the physical disk 11. (See arrows Y1, Y1 '). The C block (12C) of the physical disk 12 is a set of the A block (16A) of the physical disk 16 and the B block (11B) of the physical disk 11. (See arrows Y6, Y6 '). When a command is issued from the disk driver 22 to the A block (12A) of the physical disk 12, the mirror control unit 23 and the B block (13B) of the physical disk 13 that is paired with the corresponding A block (12A) Copying is performed on the C block (14C) of the disk 14. As a result, the data DA2 written in the A block (12A) of the physical disk 12 is copied to the B block (13B) of the physical disk 13 and the C block (14C) of the physical disk 14 (DA2 ′, DA2 '').

  The mirror controller 23 operates in the same manner for other blocks. That is, when the disk drive 10 is composed of N physical disks, the A block (16A) of the Nth physical disk (the disk indicated by reference numeral 16 in FIG. 2) is the same as that of the first physical disk 11. The B block (11B) and the C block (12C) of the second physical disk 12 are set, and the B block (16B) of the Nth physical disk 16 is the N-1th physical disk 15 The A block (15A) and the C block (11C) of the first physical disk 11 are set, and the C block (16C) of the Nth physical disk 16 is the N-2th physical disk. It is a set with 14 A blocks (14A) and B block (15B) of the (N-1) th physical disk 15. In this way, each block becomes a set (see arrows Y1 to Y6, Y1 'to Y6').

  When a command is issued from the disk driver 22 to each A block (11A, 12A, 13A, 14A, 15A, 16A), the mirror control unit 23 determines each A block based on the correspondence data stored in advance. (11A, 12A, 13A, 14A, 15A, 16A) and each B block (12B, 13B, 14B, 15B, 16B, 11B) and each C block (13C, 14C, 15C, 16C, 11C, 12C) And a mirroring process for copying the data of each A block (DA1, DA2, DA3, DA4, DA5, DA6) to each B block to be set is executed. By the mirroring process, the codes DA1 ′, DA2 ′, DA3 ′, DA4 ′, DA5 ′, DA6 ′ are written to each C block corresponding to the data of the A block. The data DA1 ″, DA2 ″, DA3 ″, DA4 ″, DA5 ″, DA6 ″ are added to the data.

  As described above, in the present invention, the disk (storage capacity) of a single physical disk in the disk array apparatus is divided into three blocks having the same capacity, and each of these blocks is designated as A block, B block, and C block. The B block of the other physical disk and the C block of a physical disk different from the other physical disk are set, the data is written to the A block, and the data written to the A block is changed to the B block and the C block. make a copy. For this reason, even if a sudden failure occurs in a physical disk, the fault tolerance performance can be improved without using a spare disk by dividing the physical disk into areas and automatically tripleting the data to different physical disks. Can be planned. At this time, since an unnecessary spare disk is not used, an increase in the number of physical disks can be suppressed, and the cost and size of the apparatus can be reduced. In particular, by making all the divided areas have the same capacity, all data is surely copied, and the reliability of the data is increased.

  Furthermore, the RAID controller 20 has a replacement recognition function for recognizing that a physical disk has been replaced. For example, the disk driver 22 constantly monitors the operating status of each of the physical disks 11 to 16 and degenerates when a failure occurs, and recognizes that a new disk has been mounted after being removed. Along with this, the mirror control unit 23 recognizes a divided area that is a set for the divided area of the physical disk that has been exchanged based on the correspondence data, and is the same as the data stored in the divided area. Has the function of writing the data in the divided area of the exchanged physical disk. Thereby, when a physical disk fails, the data is copied to the replaced divided area by replacing the disk. Thereafter, the RAID control unit 21 replaces the failed physical disk with the data copied from the failed physical disk to another physical disk and the data written to the normal physical disk based on the data recovery processing function. Restore the data needed to write to the new physical disk to be replaced.

  Further, the RAID controller 20, particularly the RAID control unit 21 and the disk driver 22 of the present disk array device 1, among the three blocks that are set based on the correspondence data from the physical disks 11 to 16 when reading data. It has a function of reading data from either. For example, the data DA1 and the like recorded in the block are read from only the A block (11A and the like) of each physical disk 11 to 16 in response to a command from the host device 30. As a result, as described above, redundancy can be maintained and an unnecessary increase in physical disks can be suppressed, and any one of the disks in the set can be accessed at the time of reading, so that the access time can be shortened. it can.

  Next, specific examples of the present invention will be described with reference to FIGS.

  Although the case where RAID is not applied has been described in the above description, FIG. 3A shows an example in which the disk array device 1 of the present invention is applied to a RAID device equivalent to the RAID 1 level.

  The disk drive 100 shown in this figure is composed of six physical disks 101 to 106, and the storage area of each physical disk is divided into three storage areas of the same capacity. As described above, on the assumption that the physical disks are different from each other, each divided area is a set with other divided areas. For example, the A block (101A) of the physical disk 101, the B block (102B) of the physical disk 102, and the C block (103C) of the physical disk 103 are set (see arrows in the figure). The same applies to other divided regions. In such a configuration, first, DA1 stored in the divided area 101A of the physical disk 101 is stored as data DA1 'in the divided area 102A of the physical disk 102 by the action of RAID1. Data DA1 'and data DA1 "are also stored in the divided area 102B of the physical disk 102 and the divided area 103C of the physical disk 103, which are paired with the divided area 101A, due to the above-described mirroring function. Further, the data DA1 ′ stored in the divided area 102A of the physical disk 102 by the action of the RAID1 is also copied to the physical disk areas 103B and 104C paired with the divided area 102A by the mirroring function (data DA1 '', Data DA1 '' '). As a result, it is possible to improve the fault tolerance performance of the disk while suppressing an unnecessary increase in the number of physical disks.

  Next, an example in which the disk array device of the present invention is applied to a RAID device equivalent to RAID 5 level will be described with reference to FIG. In the example shown in this figure, as in the case of FIG. 3A, the disk drive 100 is provided with six physical disks 101 to 106, each of which is divided into three areas (area 101A). , 101B, 101C, etc.).

  In RAID5, write data written by a command from a host device (not shown) and parity data generated corresponding to the data are distributed and stored in each data area. The B block (102B, 103B, 104B, 105B, 106B, 101B) of the other physical disk that is combined with these A blocks (101A, 102A, 103A, 104A, 105A, 106A), the other physical disk and Are copied to C blocks (103C, 104C, 105C, 106C, 101C, 102C) of another physical disk (DA1 ′ to DA6 ′, DA1 ″ to DA6 ″). This correspondence is almost the same as that shown in FIG.

  The recovery operation when a plurality of physical disks fail when applied to RAID 5 will be described with reference to FIG. First, the state shown in FIG. 4A is a case where two physical disks 102 and 103 have failed. In this case, the two failed physical disks 102 and 103 are replaced with two normal physical disks. When the physical disks 102 and 103 are replaced, the RAID controller 20 recognizes two new physical disks 102 and 103.

  Next, the RAID controller 20 reads data DA1 and data DA6 ′ from the block A of the physical disk 101, data DA6 ′ from the block B of the physical disk 104, data DA2 ″ from the block C, and copies the data DA1. Are written as data DA1 ′ and data DA1 ″ to the B block of the new physical disk 102 and the C block of the new physical disk 103, respectively (arrow YA1). Further, a copy of the data DA6 ′ is written to the C block of the new physical disk 102 as data DA6 ″ (arrow YA6), and a copy of the data DA3 ′ is written to the A block of the new physical disk 103 as data DA3 (arrow YA3). A copy of DA2 ′ is written as data DA2 and data DA2 ′ in the A block of the new physical disk 102 and the B block of the new physical disk 103, respectively (arrow YA2). The reconstruction is completed through the above processing.

  Further, as shown in FIG. 4B, when three physical disks 102, 103, and 104 fail, the three failed physical disks 102, 103, and 104 become normal three physical disks. Exchange. When the physical disks 102, 103, and 104 are exchanged, the RAID controller 20 recognizes three new physical disks 102, 103, and 104.

  Next, the RAID controller 20 reads data DA1 from the A block of the physical disk 101, reads data DA3 '' from the C block of the physical disk 105, and DA6, DA5 from the A block, B block, and C block of the physical disk 106, respectively. ', DA4' 'is read. Then, the RAID controller 20 reconstructs the lost DA2 based on the plurality of data and the parity data from the plurality of read data DA1, DA3 ″, DA6, DA5 ′, DA4 ″. This method of constructing data is a general-purpose method used for data reconstruction in RAID 5, and details thereof are omitted.

  Next, the RAID controller 20 writes the reconstructed data DA2 to the A block of the new physical disk 102. Further, the RAID controller 20 writes DA2 written to the A block of the physical disk 102 to the B block of the new physical disk 103 and the C block of the new physical disk 104. Further, the RAID controller 20 copies the data DA1 written in the A block of the physical disk 101 to the B block of the new physical disk 102 and the C block of the new physical disk 103 (arrow YB1). Further, the RAID controller 20 copies the data DA6 'written in the B block of the physical disk 101 to the C block of the new physical disk 102 (arrow YB6). Further, the RAID controller 20 copies DA3 ″ written in the C block of the physical disk 105 to the A block of the new physical disk 103 and the B block of the new physical disk 104 (arrow YB3). Further, the RAID controller 20 copies DA4 'written in the B block of the physical disk 105 to the A block of the new physical disk 104 (arrow YB4). Through the above processing, the data to be stored in the failed physical disks 102, 103, 104 is recovered.

  Further, as shown in FIG. 4C, when four physical disks 102, 103, 104, and 106 fail, the four failed physical disks 102, 103, 104, and 106 are normal. Replace with a physical disk. When the physical disks 102, 103, 104, 106 are replaced, the RAID controller 20 recognizes four new physical disks 102, 103, 104, 106.

  Next, the RAID controller 20 reads data DA1 and DA6 'from the A block and B block of the physical disk 101, and reads DA5, DA4' and DA3 "from the A block, B block and C block of the physical disk 105, respectively. Then, the RAID controller 20 reconstructs the lost DA2 based on the plurality of data and parity data from the plurality of read data DA1, DA6 ', DA5, DA4', DA3 ". This method of constructing data is a general-purpose method used for data reconstruction in RAID 5, and details thereof are omitted.

  Next, the RAID controller 20 writes the reconstructed data DA2 to the A block of the new physical disk 102. Further, the RAID controller 20 writes DA2 written to the A block of the physical disk 102 to the B block of the new physical disk 103 and the C block of the new physical disk 104. Further, the RAID controller 20 copies the data DA1 written in the A block of the physical disk 101 to the B block of the new physical disk 102 and the C block of the new physical disk 103 (arrow YC1). Further, the RAID controller 20 copies the data DA6 'written in the B block of the physical disk 101 to the C block of the new physical disk 102 and the A block of the new physical disk 106 (arrow YC6). Further, the RAID controller 20 copies the data DA3 ″ written in the C block of the physical disk 105 to the A block of the new physical disk 103 and the B block of the new physical disk 104 (arrow YC3). Further, the RAID controller 20 copies DA4 'written in the B block of the physical disk 105 to the A block of the new physical disk 104 and the C block of the new physical disk 106 (arrow YC4). Further, the RAID controller 20 copies DA5 written in the A block of the physical disk 105 to the B block of the new physical disk 106 (arrow YC5). Through the above processing, the data to be stored in the failed physical disks 102, 103, 104, 106 is recovered.

  As described above, the disk array device 1 according to the present invention has a high data redundancy rate and suppresses an increase in the number of disks even when many disks have failed, thereby achieving downsizing and cost reduction. be able to. For example, it is compared with the disk array device in the conventional example of the mirror / RAID combination method described in Patent Document 1 described above. First, assuming that the capacity per physical disk of the present application is X, as shown in FIG. 5, the capacity per one physical disk is 1/3, and the storage capacity of the entire HDD is the same. . For example, a case where two physical disks have failed is compared. In the present application, even if 2 out of 6 units fail, since the data is tripled, the operation can be continued without degeneration by 100%, and the recovery process can be executed in a redundant state. On the other hand, in the case of the conventional example, when two units fail in the configuration of 6 sets and 12 units, the setback occurs when two sets fail, so the recovery process must be performed with a probability of about 9% while maintaining the reduced state. . If there is no failure, the number of physical disks can be halved despite the fact that the entire actual data use capacity is the same, so the cost of the physical disks can be reduced. Furthermore, as the number of configurations increases, the space for storing the physical disk is halved, so that the redundant disk housing is halved and the cost can be reduced.

  Further, the disk array device in the conventional example of the redundant disk system described in Patent Document 2 described above and the disk array device in the present invention will be compared. The storage capacity per physical disk in the present invention is X, and the storage capacity per physical disk of Patent Document 2 is assumed to be 1/3. Under this condition, the total storage capacity in the present invention is 6X in the case of six physical disks. The total storage capacity in the prior art is reduced to (11/3) X in total for data and redundancy when there are 11 physical disks. Here, if the physical disk cost is, for example, (capacity X: capacity 1/3 · X) = (36:17), the total disk cost is 216: 187. However, the larger the configuration of the number, the greater the number of redundant disks in the prior art and the greater the number of housings to be mounted. Capacity X HDD: Housing to be mounted (for mounting 15 HDDs) = 4: 5 (Capacity X: Capacity 1/3 · X: Mounting chassis) = (36:17:40) It becomes. When the number of HDDs is doubled as described above, 12 HDDs with a capacity X used in the present invention, one chassis, and 24 HDDs with a capacity of 1/3 · X in the case of the redundant disk system of the prior art, mounted The cost is 472: 488, which is reversed. That is, the larger the configuration, the lower the cost of the present invention.

  As described above, in the present invention, one physical disk shares a plurality of data, so that data recorded in one physical disk is formed on different physical disks for each divided area. Therefore, even if a plurality of physical disks fail, data can be easily recovered and data redundancy can be maintained. In addition, an increase in the number of physical disks can be suppressed, and the cost and size of the apparatus can be reduced. In particular, by making all the divided areas have the same capacity, all the data is surely copied, and the reliability of the data is increased.

  In the above-described embodiment, the case where the present invention is applied to RAID 1 and RAID 5 has been described. However, there are RAID 0, 2, 3, and 4 schemes in addition to this.

  The RAID 0 method is a method (striping) in which data is divided into block units and distributed over a plurality of physical disks to record the data (striping). The RAID 2 system uses ECC (Error Correction Code: also called “Haming code” or “error correction code”) used in a main storage device or the like to detect and correct data errors. In this method, data is divided into bytes and stored on a plurality of physical disks dedicated to data. The RAID3 method divides data into block units, bits or bytes, and writes simultaneously to a plurality of physical disks dedicated to data. Parity is generated from the divided data and written to the physical disk dedicated to parity. This is a method of accessing a physical disk in parallel and transferring data in a batch. The RAID4 method is a striping of RAID0 and has a function of regenerating data by adding a physical disk dedicated to parity. The data is divided into block units and recorded on the data physical disk, and the parity is 1 Is recorded on a physical disk dedicated to parity. The present invention can be similarly applied to these RAID systems.

1 is a schematic block diagram of a disk array device of the present invention. It is a schematic block diagram of the disk drive in FIG. FIG. 3A is a diagram illustrating an example applied to a RAID device corresponding to a RAID 1 level, and FIG. 3B is a diagram illustrating an example applied to a RAID device corresponding to a RAID 5 level. 4A is a diagram for explaining a recovery operation when two physical disks fail when RAID 5 is applied, and FIG. 4B is a diagram when three physical disks fail. FIG. 4C is a diagram for explaining the restoration operation, and FIG. 4C is a diagram for explaining the restoration operation when four physical disks fail. It is a schematic block diagram which shows a prior art. It is a schematic block diagram which shows a prior art.

Explanation of symbols

1, 200, 400 Disk array device 10, 100 Disk drive 11-16 Physical disk 101-106 Physical disk 20, 250, 450 RAID controller 21 RAID control unit 22 Disk driver 23 Mirror control unit 30, 300, 500 Host device 410 Data Disk drive 420 Redundant disk drive for row direction 430 Redundant disk drive for row direction

Claims (6)

A plurality of physical disks and a control unit that reads / writes data from / to the storage area of the physical disk based on a command from a higher-level device, and each of the physical disks has the same storage capacity The three blocks are divided into a first block, a second block, and a third block. The first block is a second block of another physical disk having the same capacity, and a third block different from the two physical disks. A disk array device characterized in that a block is set in a set relation. The set relation is as follows: one divided storage area of the one physical disk, one divided storage area of the other physical disk, and one divided physical disk different from the other physical disk. The other physical storage area, the other physical storage area divided from the one physical disk, the other storage area of the other physical disk, and the physical disk different from the other physical disk. The other storage area of the one physical disk, the other divided storage area of the other physical disk, and the other physical disk are further separated from each other. 2. The disk array device according to claim 1, wherein the disk array device is set between one of the divided storage areas of the physical disk. The control unit has correspondence data for associating the storage areas to be the set with each other, and has a management function for holding the same data in the storage areas to be the set based on the correspondence relation data. The disk array device according to claim 1. The storage area to be set is set as an area for writing data, and the control unit has write area specifying data for specifying the storage area as a data writing area, and responds to a command from a host device. A function of copying the same data as the data written in the storage area to another storage area set as the storage area based on the write area specifying data. The disk array device according to claim 1. The disk array device according to claim 3, wherein the control unit has a function of reading write data from any one of the storage areas to be set based on the correspondence data. The control unit has an exchange recognition function for recognizing the exchange of the physical disk, recognizes the exchange of the physical disk, and exchanges data in storage areas of other physical disks in the set relation. 2. The disk array device according to claim 1, further comprising a function of copying to a storage area of the physical disk.

Images