JP2007087062A - Array type storage device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an array type storage device, in which data writing speed can be improved in writing data to a storage device array comprising a RAID 4 or RAID 5. <P>SOLUTION: An array controller 20 refers to a data storage status table storing data storage status information showing storage of data for each block in association with each of stripes containing a plurality of blocks configured on a disk array 300, and assigns a block contained in a stripe in which data is not stored in all blocks as a storage area for writing data. The array controller 200 generates parity data corresponding to the writing data, and assigns, as an area for the parity data, another block of a stripe containing a block assigned as an area for writing the writing data. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to an array type storage device that has a storage device array in which a plurality of storage devices are arrayed and controls access to the storage device array in accordance with an instruction received from an external device.

  2. Description of the Related Art In recent years, large-capacity multimedia data such as moving images and still images have been handled with the improvement of processing capability of information devices and the widening of communication networks. As a storage suitable for storing such multimedia data, a storage having high speed, large capacity, and high reliability is required. In particular, the market dealing with high-definition video distribution via broadband networks is expected to rapidly expand in the future, and it is an urgent need to provide a storage solution suitable for storing multimedia data.

  As one of technologies for realizing high-speed, large-capacity, and high-reliability storage, RAID proposed by D. Patterson and the like is widely known (for example, see Non-Patent Document 1).

  In the above-mentioned document, data is divided into a plurality of blocks at RAID level 4 (hereinafter referred to as RAID 4) or RAID level 5 (hereinafter referred to as RAID 5), and parity data is generated from the data divided into the plurality of blocks. A method of storing data efficiently in terms of capacity while ensuring redundancy is proposed by distributing and storing the data divided into the plurality of blocks and the parity data in a plurality of disk devices. .

  When new data is written to a disk array configured with RAID 4 or RAID 5, a read-modify-write operation is performed. That is, (1) the old data currently recorded in the new data write destination block and the old parity data currently recorded in the parity block corresponding to the new data write destination block are read. (2) The new parity data is calculated by taking an exclusive OR of the new data, the old data, and the old parity data. (3) The operation of writing new data and new parity data in the respective write destination blocks is performed. That is, in order to write new data, two reading processes are required.

Further, a disk array device for improving the data writing speed has been proposed (see Patent Document 1). This disk array device has a use state storage unit for storing the use state of each block constituting the disk array. When writing data, the disk array device refers to the use state storage unit, and when the data write destination block is unused, the read processing from the data write destination block is omitted and the parity data is stored. Generate. As a result, the write penalty is reduced and the data writing speed is improved.
JP-A-6-75708 D. Patterson et al. A Case for Redundant Arrays of Inexpensive Disks (RAID), inACM, SIGMOD Conference, Chicago, IL, June 1988

  However, in the above-described conventional technology, when data is written to a disk array configured with RAID 4 or RAID 5, a so-called write penalty occurs in which the writing speed is remarkably reduced due to a read-modify-write operation.

  In the disk array device described above, when the data write destination block is unused, the write penalty is reduced by generating the parity data by omitting the read process from the data write destination block. . However, when data is already stored in the data write destination block, or when data is already stored in at least one block of the stripe including the data write destination block, a write penalty still occurs. Therefore, this disk array device cannot improve the data writing speed in all cases.

  An object of the present invention is to provide an array type storage device capable of improving the data writing speed in data writing to a storage device array constituted by RAID4 or RAID5.

  In order to achieve the above object, the present invention has an array type storage that has a storage device array in which a plurality of storage devices are arrayed and controls access to the storage device array in accordance with an instruction received from an external device. Data storage status information for storing data storage status information indicating presence or absence of data storage for each storage unit in association with each of redundancy groups including a plurality of storage units configured on the storage device array A storage area that refers to the storage means and the data storage state information storage means, and performs storage unit allocation as a storage area for write data and storage unit allocation as a redundant data storage area corresponding to the write data Allocating means, redundant data generating means for generating redundant data corresponding to the write data, the write data and the redundant data For each of the storage units allocated by the corresponding storage area allocation unit, the storage area allocation unit as a storage area for the write data in a redundant group in which no data is stored in all the storage units. Provided is an array type storage device in which a storage unit is allocated and another storage unit including a storage unit assigned as an area for writing the write data is assigned as an area for writing the redundant data. .

  According to the present invention, a read-modify-write operation does not occur when data is written to a storage device array configured by RAID 4 or RAID 5, the occurrence of a write penalty can be eliminated, and the data writing speed can be improved. it can.

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

(First embodiment)
FIG. 1 is a block diagram showing the configuration of the disk array device according to the first embodiment of the present invention.

  As shown in FIG. 1, the disk array device includes a host device 100, an array controller 200, and a disk array (storage device array) 300. The host device 100 includes a CPU 101, a ROM 102, a RAM 103, and a storage device I / F (interface) circuit 104. The array controller 200 includes an MPU 201, a ROM 202, a RAM 203, a parity operation circuit 204, a host device I / F circuit 205, and hard disk drive I / F circuits (HDD I / F circuits) 206 to 209. The disk array 300 includes a plurality of hard disk drives (hereinafter referred to as HDDs) 301 to 304 configured to be able to implement RAID4 or RAID5.

  In the present embodiment, a description will be given of an embodiment in which the storage device constituting the disk array is an HDD, but the storage device constituting the disk array is not limited to the HDD. Other types of storage devices such as optical disk drives, magneto-optical disk drives, and semiconductor storage devices may be used.

  The host apparatus 100 issues to the array controller 200 an instruction related to the configuration of the disk array 300 and an instruction related to writing / reading data to / from the disk array 300. Also, write data to the disk array 300 is transferred to the array controller 200.

  When the array controller 200 receives a command from the host device 100, the array controller 200 interprets the command and configures the disk array 300 and writes / reads data to / from the disk array 300. When the array controller 200 receives a data write command and write data from the host device 100 to the disk array 300, the array controller 200 reconfigures the received write data based on the configuration of the disk array 300, and performs the reconfiguration on the reconfigured data. To generate parity data. Then, the array controller 200 issues a data write command to some or all of the HDDs 301 to 304 constituting the disk array 300, and writes the reconfigured data and the parity data in a distributed manner.

  When the array controller 200 receives a data read command from the disk array 300 from the host device 100, the array controller 200 issues a data read command to some or all of the HDDs 301 to 304 constituting the disk array 300 to read data. The read data is reconfigured based on the configuration of the disk array 300 and transferred to the host device 100.

  Upon receiving a data write command and write data from the array controller 200, each of the HDDs 301 to 304 records the received write data on a magnetic disk recording medium. Further, upon receiving a data read command from the array controller 200, each of the HDDs 301 to 304 reproduces data from the magnetic disk recording medium, and transfers the reproduced data to the array controller 200.

  Next, a characteristic operation of the disk array device according to the present embodiment will be described with reference to FIG. FIG. 2 is a flowchart showing an operation procedure of the array controller 200 when writing data to the disk array 300 in the disk array apparatus of FIG. The procedure shown in the flowchart of FIG. 2 is executed by the MPU 201 of the array controller 200.

  As shown in FIG. 2, when the array controller 200 receives a data write command and write data from the host device 100 in step S201, the array controller 200 reconfigures the received write data based on the configuration of the disk array 300 in step S202. To do. Then, a block on the disk array 300 is assigned to the reconstructed data and the parity data generated from the reconstructed data by a method described later, and parity data is generated from the reconstructed data. In step S203, the array controller 200 writes the reconstructed data and the parity data in the allocated block.

  Next, a method for allocating blocks on the disk array 300 to the reconstructed data and the parity data and generating the parity data will be described with reference to FIGS. 3 is a flowchart showing the procedure of block allocation / parity data generation processing in step S202 of FIG. 2, and FIG. 4 is a disk array at the time of formatting, a data storage state table, and an LBA-disk array block conversion table. FIG. 5 shows a disk array, a data storage state table, and an LBA-disk array when a write command for data D0 to D1 having a size of 2 blocks is issued from the host device to LBA = L0 to L1. FIG. 6 is a diagram showing each state of the block conversion table. FIG. 6 shows a disk array when a write command for data D2 having a size of one block from the host device to LBA = L2 is issued in the state of FIG. , Data storage state table, and LBA-disk array block FIG. 7 is a diagram showing the state of each conversion table. FIG. 7 shows a state where a write command for data D3 to D6 having a size of 4 blocks from the host device to LBA = L3 to L6 is issued in the state of FIG. FIG. 8 is a diagram showing the status of each of the disk array, data storage status table, and LBA-disk array block conversion table, and FIG. 8 is the size of one block from the host device to LBA = L2 in the status of FIG. FIG. 7 is a diagram showing a state of each of a disk array, a data storage state table, and an LBA-disk array / block conversion table when a write command for data D7 having a data is issued. Here, the disk array 300 will be described as being configured by RAID 5, but the present invention is applicable even if it is configured by RAID 4. Further, the number of HDDs is not limited to four.

  When the array controller 200 is turned on, the data storage state table (hereinafter referred to as storage state table) 203a and the LBA-disk array block conversion table (hereinafter referred to as conversion table) as shown in FIGS. 203b is read from the HDDs 301 to 304 or other non-volatile storage media and expanded on the RAM 203. The storage state table 203a is a bitmap table that holds information indicating whether data is stored in a block. The conversion table 203b is a table that holds information indicating a correspondence between a logical address (hereinafter referred to as LBA) recognized by the host apparatus 100 and a block position on the disk array 300.

  As shown in FIG. 3, first, the array controller 200 refers to the storage state table 203a in step S301 to search for a stripe in which no data is stored in any block. In subsequent step S302, the storage state table 203a and the conversion table 203b are updated to allocate blocks on the disk array 300 to the reconstructed data and the parity data. Here, the storage state table 203a and the conversion table 203b updated on the RAM 203 are copied to the HDDs 301 to 304 or other nonvolatile storage medium at an arbitrary timing, for example, when the power is shut off. In step S303, the array controller 200 generates the parity data from the data reconstructed by the parity calculation circuit 204.

  Specifically, when the disk array 300 is formatted, the state shown in FIG. Since this state is a state in which no data is stored in the disk array 300, the storage state table 203a stores data for all bits corresponding to each block, as shown in FIG. “0” indicating that the data is not stored is written. Further, as shown in FIG. 4C, valid information is not written in the conversion table 203b.

  In the formatted state of FIG. 4, when a write command for data D0 to D1 having a size of two blocks to LBA = L0 to L1 is issued from the host device 100, the disk array is transferred to FIG. It will be in the state shown in. Here, in order to facilitate understanding of the present invention, the data cluster size managed by the host apparatus 100 and the block size of the disk array 300 are described as being equal, but of course, they may be different. Further, the formats of the storage state table 203a and the conversion table 203b are not limited to those shown in the figure. Furthermore, the configuration of the table is not limited to that illustrated.

  When the array controller 200 receives the write command, the array controller 200 refers to the storage state table 203a to find the stripe S0 as a stripe in which data for two blocks can be stored and no data is stored in any block. In the stripe S0, the HDDs 301 to 302 are selected as areas for two blocks. Here, for the parity data P0 for the stripe S0, the HDD 304 is automatically selected according to the RAID5 format. Next, bits (strip number, HDD number) = (S0, 301) and (S0, 302) corresponding to the selected area on the storage state table 203a are rewritten to “1”. Also, for the conversion table 203b, the correspondence between the LBA recognized by the host device 100 and the block position on the disk array 300, that is, (LBA, HDD number, stripe number) = (L0, 301, S0), (L1, 302, S0) is added. In this way, blocks of the disk array 300 are allocated to the data D0 to D1 and the parity data P0 received from the host device 100.

Next, the parity data P0 is P0 = D0 xor D1
The generated parity data P0 is written together with the data D0 and D1 to the blocks assigned to them.

  In the state shown in FIG. 5A, when a write command for data D2 having a size of one block to LBA = L2 is issued from the host device 100, the disk array 300 is transferred to FIG. 6A. It will be in the state shown.

  When the array controller 200 receives the write command, the array controller 200 refers to the storage state table 203a and searches for the stripe S1 as a stripe in which data for one block can be stored and no data is stored in any block. In the stripe S1, the HDD 301 is selected as an area for one block. For the parity data P1 for the stripe S1, the HDD 303 is automatically selected according to the RAID5 format. Then, as shown in FIG. 5B, the bit (stripe number, HDD number) = (S1, 301) corresponding to the selected area on the storage state table 203a is rewritten to “1”. Further, as shown in FIG. 5C, for the conversion table 203b, the correspondence between the LBA recognized by the host device 100 and the block position on the disk array 300, that is, (LBA, HDD number, stripe number) = ( L2, 301, S1) is added. In this way, blocks of the disk array 300 are allocated to the data D2 and parity data P1 received from the host device 100.

Next, the parity data P1 is
P1 = D2
The generated parity data P1 is written to the blocks allocated to the data D2 together with the data D2.

  In the state of FIG. 6, when a write command for data D3 to D6 having a size of 4 blocks from the host device 100 to LBA = L3 to L6 is issued, the disk array 300 is in the state shown in FIG. become.

  Upon receiving the write command, the array controller 200 refers to the storage state table 203a and searches for the stripes S2 and S3 as stripes that can store data for four blocks and no data is stored in any block. . The array controller 200 selects the HDD 301 and the HDDs 303 and 304 as an area for three blocks in the stripe S2, and selects the HDD 302 as an area for one block in the stripe S3. Here, for the parity data P2 for the stripe S2, the HDD 302 is automatically selected according to the RAID5 format. For the parity data P3 for the stripe S3, the HDD 301 is automatically selected according to the RAID5 format.

  Then, as shown in FIG. 7B, bits (stripes number, HDD number) corresponding to the selected area on the storage state table 203a = (S2, 301), (S2, 303), (S2, 304). And (S3, 302) are rewritten to “1”. Further, as shown in FIG. 7C, the correspondence between the LBA recognized by the host apparatus 100 and the block position on the disk array 300 is added to the conversion table 203b. That is, (LBA, HDD number, stripe number) = (L3, 301, S2), (L4, 303, S2), (L5, 304, S2) and (L6, 302, S3) are added. In this way, blocks of the disk array 300 are assigned to the data D3 to D6 and the parity data P2 and P3 received from the host device 100.

After that, the parity data P2 and P3 are respectively P2 = D3xorD4xorD5
P3 = D6
The generated parity data P2 and P3 are written in the blocks assigned to the data D3 to D6.

  In the state shown in FIG. 7, when the host device 100 issues a write instruction for data D7 having a size of one block to LBA = L2, that is, an overwrite instruction to LBA = L2, the disk array 300 is changed to FIG. The state shown in FIG.

  In this case, the array controller 200 is the same as in the case of FIGS. That is, when the write command is received, the storage state table 203a is referred to search for the stripe S4 as a stripe in which data for one block can be stored and no data is stored in any block. Then, the HDD 301 is selected as an area for one block in the stripe S4. Here, for the parity data P4 for the stripe S4, the HDD 304 is automatically selected according to the RAID5 format. Then, as shown in FIG. 8B, the bit (stripe number, HDD number) = (S4, 301) corresponding to the selected area on the storage state table 203a is rewritten to “1”. Further, as shown in FIG. 8C, the correspondence between LBA = L2 recognized by the host apparatus 100 on the conversion table 203b and the block position on the disk array 300 is (LBA, HDD number, stripe number) = (L2 , 301, S4). In this way, blocks of the disk array 300 are allocated to the data D7 and parity data P4 received from the host device 100.

Next, the parity data P4 is P4 = D7
The generated parity data P4 is written together with the data D7 to the blocks assigned to each.

  In this embodiment, the array controller 200 manages the LBA when the host device 100 issues an overwrite command. Instead of this, the host device 100 manages the LBA by, for example, an operating system or a device driver so as not to specify an LBA in which data is already stored in a data write command to the disk array 300. It may be adopted.

  Next, the operation of the array controller 200 when reading data from the disk array 300 into which data has been written in the above-described procedure will be described with reference to FIG. FIG. 9 is a flowchart showing an operation procedure of the array controller 200 when reading data from the disk array 300. The procedure shown in the flowchart of FIG. 9 is executed by the MPU 201.

  As shown in FIG. 9, the array controller 200 receives a data read command from the host device 100 in step S901. Subsequently, in step S902, the array controller 200 refers to the conversion table 203b and searches for a block on the disk array 300 corresponding to the LBA designated by the host device 100. Next, the array controller 200 issues a data read command to the HDD including the block corresponding to the LBA in step S903. In step S904, the array controller 200 determines whether reading of the designated data has failed. If the reading of the designated data is successful, the array controller 200 transfers the read data to the host device 100 in step S906.

  On the other hand, when it is determined in step S904 that reading of the designated data has failed, the array controller 200 proceeds to step S905. In this step, the array controller 200 restores the data that has failed to be read from other data and parity data on the same stripe by a method described later. In step S906, the array controller 200 transfers the restored data to the host device 100.

  Here, a method for restoring data that has failed to be read from other data and parity data on the same stripe will be described with reference to FIG. FIG. 10 is a flowchart showing the procedure of the data restoration process in step S905 of FIG.

  When performing the data restoration processing, as shown in FIG. 10, the array controller 200 refers to the storage state table 203a in step S1001, and stores data on the same stripe as the data that failed to be read. Search for a block. In step S1002, the array controller 200 determines whether a block storing data is found. If a block in which data is stored is found, the process proceeds to step S1003. In step S1003, the array controller 200 issues a data read command to the HDD including the block storing the data and the HDD including the block storing the parity data corresponding to the data that has failed to be read. Then, the data and parity data stored in the block on the same stripe as the data that has failed to be read are read. In step S1004, the array controller 200 restores the data that failed to be read by taking an exclusive OR of all the read data.

  On the other hand, if it is determined in step S1002 that a block storing data is not found, the process proceeds to step S1005. In step S1005, the array controller 200 issues a data read command only to the HDD including the block storing the parity data corresponding to the data that failed to be read, and only the parity data corresponding to the data that failed to be read. Is read. This is because the parity data corresponding to the data that failed to be read has the same value as the data that failed to be read. Then, the array controller 200 restores the data that failed to be read by setting the read parity data as data that failed to be read.

For example, in the state of FIG. 7, when reading of data D0 fails, data D1 and parity data P0 on the same stripe S0 as data D0 are read,
D0 = D1 xor P0
Is calculated. By this calculation, the data D0 that has failed to be read is restored.

Further, when the reading of the data D2 fails, since the data D2 is the only data stored on the same stripe S1, the parity data P1 is read. And
D2 = P1
Thus, the data D2 is restored.

  As a data restoration process that replaces the data restoration process described with reference to FIG. 10, the data restoration process shown in the flowchart of FIG. 11 may be employed. This alternative data restoration process will be described with reference to FIG. FIG. 11 is a flowchart showing the procedure of another data restoration process instead of the data restoration process of FIG.

  For the disk array 300 to which this alternative data restoration processing is applied, “0” is written in all blocks used as RAID 4 or RAID 5 at the time of formatting. In this alternative data restoration process, as shown in FIG. 11, the array controller 200 first issues a data read command to all remaining HDDs except for the HDD including the block that stores the data that failed to be read in step S1101. Issue. As a result, the values stored in all other blocks on the same stripe as the block storing the data for which the read failed are read. In step S1102, the array controller 200 restores data that has failed to be read by calculating an exclusive OR of the values read from all the other blocks. That is, in this data restoration process, as in the data restoration process shown in FIG. 10, valid data is stored on the same stripe as the block in which the data that has failed to be read is stored with reference to the storage state table 203a. The data that has failed to be read can be restored without performing the process of searching for the block being read.

For example, in the state of FIG. 7, when reading of data D0 fails, the values stored in all other blocks on the same stripe S0 as the block storing data D0, that is, data D1, on HDD 303 "0" and parity data P0 are read out,
D0 = D1 xor '0' xor P0
(The result of this calculation is the same value as D1 xor P0). By this calculation, data D0 that has failed to be read is restored.

Similarly, when the reading of the data D2 fails, the values stored in all other blocks on the same stripe S1 as the block storing the data D2, that is, “0” on the HDD 302, HDD 303 The upper “0” and parity data P1 are read,
D2 = '0' xor '0' xor P1
(The result of this calculation is the same value as P1). As a result, the data D2 that has failed to be read is restored.

  As described above, according to the present embodiment, a read-modify-write operation does not occur in writing data to the disk array 300 configured with RAID 4 or RAID 5, and the write penalty is eliminated. Thereby, the data writing speed can be improved.

(Second Embodiment)
Next, a second embodiment of the present invention will be described with reference to FIGS. FIG. 12 is a flowchart showing a procedure of block allocation / parity data generation processing in the disk array device according to the second embodiment of the present invention. FIG. 13 is a diagram showing the states of the disk array, the data storage state table, and the LBA-disk array / block conversion table at the time of data writing in the second embodiment of the present invention.

  In the present embodiment, block allocation / parity data generation processing different from that of the first embodiment is performed. Here, in FIG. 12, the steps denoted by the same reference numerals as those in FIG. 3 mean that the same processing as the steps having the same reference numerals in FIG. 3 is performed.

  In the block allocation process, the array controller 200 first analyzes the data write command received from the host device 100 in step S1201. In a succeeding step S1202, it is determined whether or not the data write command is a streaming command. Here, when the data write command is a streaming command, the same block allocation / parity data generation processing as in the first embodiment, that is, steps S301 to S303 are executed.

  Here, the streaming instruction means an instruction relating to time critical data transfer rather than a standard data integrity critical instruction. For example, the streaming command is a command that guarantees a stream request in an AV (Audio Visual / Audio Video) type application of the HDD. More specifically, in ATA / ATAPI-7 (AT Attachiment with Packet Interface-7) Working Draft Revision 3c, there are instructions defined as WRITE STREAM PIO and WRITE STREAM DMA in the streaming feature set.

  If it is determined in step S1202 that the data write command is not a streaming command, the array controller 200 allocates blocks as conventionally performed in a disk array configured with RAID 4 or RAID 5, Generate parity data. That is, the array controller 200 searches for a block corresponding to the LBA designated by the host device 100 with reference to the conversion table 203b in steps S1203 to S1204.

  However, if there is no block corresponding to the LBA designated by the host device 100 (step S1204), the array controller 200 searches for a block in which no data is stored with reference to the storage state table 203a in step S1205. In step S302, the array controller 200 updates the storage state table 203a and the conversion table 203b, thereby allocating blocks on the disk array 300 to data reconfigured based on the configuration of the disk array 300.

  Next, if necessary, the array controller 200 reads the old data stored in the write destination block and the old parity data corresponding to the old data from the HDDs 301 to 304 in the same manner as in the prior art in step S1206. Generate new parity data.

  For example, in the state of FIG. 7, when the host device 100 issues a write instruction for data D7 having a size of one block to LBA = L2, that is, an overwrite instruction to LBA = L2, the disk array 300 It will be in the state shown to 13 (a).

  Upon receiving the write command, the array controller 200 analyzes the write command and determines whether or not the write command is a streaming command. Here, when the write command is a streaming command, the array controller 200 operates in the same manner as described with reference to FIG. 8 in the first embodiment.

On the other hand, when the write command is not a streaming command, the array controller 200 refers to the conversion table 203b and searches for (HDD number, stripe number) = (301, S1) as a block corresponding to LBA = L2. . Subsequently, the data D0 stored in the block and the parity data P1 corresponding to the block are read from the HDD 301 and the HDD 303, respectively, and new parity data P1 ′ is P1 ′ = D0 xor D7 xor P1.
Is generated based on Then, the data D7 is written to the HDD 301 and the parity data P0 ′ is written to the HDD 304.

  In the present embodiment, every time the array controller 200 receives a data write command from the host device 100, it determines the attribute of the data write command and selects a block allocation / parity generation method. Not limited to this, the host device 100 prepares an instruction for selecting the block allocation / parity generation method of the array controller 200, and before issuing a data write command to the array controller 200, the block allocation / parity generation method selection instruction in advance. Can be adopted to select the block allocation / parity generation method of the array controller 200.

  As described above, according to the present embodiment, when data is written to the disk array 300 configured with RAID 4 or RAID 5, the write operation is changed based on the attribute of the data write command issued by the host device 100. The data writing speed can be improved and the capacity of the disk array 300 can be used efficiently.

(Third embodiment)
Next, a third embodiment of the present invention will be described with reference to FIG. FIG. 14 is a flowchart showing a procedure of block allocation / parity data generation processing in the disk array device according to the third embodiment of the present invention.

  In the present embodiment, block allocation / parity generation method selection processing different from that of the second embodiment is performed. Therefore, in the present embodiment, a description will be given centering on differences from the second embodiment. Here, in FIG. 14, the same steps as those in FIGS. 3 and 12 are denoted by the same reference numerals.

  In the block allocation process, the array controller 200 first analyzes the data write command received from the host device 100 in step S1401, as shown in FIG. Subsequently, in step S1402, the array controller 200 determines whether or not the data write destination LBA specified by the data write command is a high-speed write area. Here, when the data write destination LBA designated by the data write instruction is a high-speed write area, the same block allocation / parity data generation processing as in the first embodiment, that is, steps S301 to S303 are executed.

  However, here, the host apparatus 100 is assumed to know that the disk array 300 is divided into a high-speed data writing area and a normal data writing area, and one area can be selected by specifying an LBA. However, in practice, the disk array 300 does not have to be physically divided in advance as described above.

  If it is determined in step S1402 that the data write destination LBA specified by the data write instruction is not a normal write area, steps S1203 to S1206 described in the second embodiment are executed.

  As described above, according to the present embodiment, in the data write to the disk array 300 configured with RAID 4 or RAID 5, the write operation is performed based on the data write destination LBA designated by the data write command issued by the host device 100. change. As a result, the data writing speed can be improved and the capacity of the disk array 300 can be used efficiently.

(Fourth embodiment)
Next, a fourth embodiment of the present invention will be described with reference to FIGS. FIG. 15 is a flowchart showing a procedure of data erasure processing in the disk array device according to the fourth embodiment of the present invention. FIG. 16 is a diagram showing the states of the disk array, the data storage state table, and the LBA-disk array / block conversion table before execution of the data erasing process according to the fourth embodiment of the present invention. FIG. 17 is a diagram showing the states of the disk array, the data storage state table, and the LBA-disk array / block conversion table after execution of the data erasing process according to the fourth embodiment of the present invention.

  In the present embodiment, as in the first to third embodiments, processing for erasing data that has exceeded a predetermined storage time among data written to the disk array 300 is performed.

  When erasing data that has exceeded a predetermined storage time from data written to the disk array 300, the array controller 200 refers to the storage state table 203c in step S1501, as shown in FIG. Here, in the storage state table 203c, for each stripe, the time when the latest data is written in the blocks constituting the stripe is recorded. The latest data write time is measured by a timer circuit (not shown) in the array controller 200 and updated when data is written to the blocks constituting the stripe.

  Next, the array controller 200 searches for a stripe in which data is stored in steps S1502 to S1503 and steps S1507 to S1508. If there is a stripe in which data is stored, the array controller 200 compares the latest data write time of the stripe with a predetermined data storage time in steps S1504 to S1505, and saves according to the comparison result. Determine whether the deadline has been exceeded. Here, if the latest data write time indicates a time before the data storage time, the array controller 200 determines that the storage time limit has been exceeded. The data storage time is stored in, for example, the HDDs 301 to 304 or other non-volatile storage media, and is called on the RAM 203 during the data erasing process.

  If the storage time limit has been exceeded, the array controller 200 updates the storage state table 203c and the conversion table 203b in step S1506, thereby confirming the presence of data on the stripe from the storage state table 203c and the conversion table 203b. to erase. Further, for example, the stripe may be initialized by writing “0” in all the blocks constituting the stripe.

  Next, for example, assuming that data is written in the disk array 300 as shown in FIG. 16, a case will be described in which data that has exceeded a predetermined storage time is erased. Here, the data storage time is 50 days.

  The array controller 200 refers to the storage state table 203c and first searches for the stripe S0 as the stripe in which data is stored. Regarding this stripe S0, the latest data write time 40 days 4 hours 18 minutes (indicated as 40:04:18) indicates the time before the data storage time 50 days, so the column of the stripe number S0 in the storage state table 203c Is returned to the initial state (see FIG. 4) (see FIG. 17). Also, the column in which the stripe number S0 is recorded on the conversion table 203b, that is, (LBA, HDD number, stripe number) = (L0, 301, S0), (L1, 302S0) is deleted (see FIG. 17). ).

  Next, the stripe S1 is found as a stripe in which data is stored. Regarding the stripe S1, since the latest data writing time 40 days 4 hours 20 minutes indicates the time before the predetermined data storage time 50 days, the portion relating to the stripe S1 in the storage state table 302c is returned to the initial state (FIG. 17). See). However, since there is no column in which the stripe number S1 is recorded in the conversion table 203b, the conversion table 203b is not rewritten (see FIG. 17).

  Next, the stripe S2 is found as a stripe in which data is stored. Regarding the stripe S2, since the latest data writing time 51 days 12 hours 8 minutes indicates the time after the predetermined data storage deadline 50 days, the storage state table 302c and the conversion table 203b are not rewritten (see FIG. 17). reference). Thereafter, similarly, a stripe in which data is stored is found and the same processing is performed.

  The data erasing process may be performed at a predetermined timing when the array controller 200 is not accessed from the host device 100, for example, when a predetermined time elapses or when the power is turned on. Alternatively, the host apparatus 100 may prepare an instruction for causing the array controller 200 to perform the data erasure process, and may be performed when the array controller 200 receives the data erasure instruction from the host apparatus 100.

  As described above, according to the present embodiment, in the disk array 300 configured with RAID 4 or RAID 5, among the data written to the disk array 300 as in the first to third embodiments, a predetermined number of data is written. Erase data that exceeds the storage time. As a result, the capacity of the disk array 300 can be used efficiently.

(Fifth embodiment)
Next, a fifth embodiment of the present invention will be described with reference to FIGS. FIG. 18 is a flowchart showing a procedure of data erasure processing in the disk array device according to the fifth embodiment of the present invention. FIG. 19 is a diagram showing a state of each of the disk array, the data storage state table, and the LBA-disk array / block conversion table after execution of the data erasing process according to the fifth embodiment of the present invention.

  In this embodiment, the data stored in the unallocated stripe is erased from the data written to the disk array 300 as in the first to third embodiments, compared to the fourth embodiment. It is different in point to do.

  As shown in FIG. 18, the array controller 200 refers to the conversion table 203b in step S1801. Subsequently, in step S1802, the array controller 200 extracts an unallocated stripe, that is, a stripe to which any block of the LBA recognized by the host device 100 is not allocated.

  Next, in steps S1803 and S1804, the array controller 200 refers to the storage state table 203c for the extracted unallocated stripe and determines whether data is stored. In step S1805, the array controller 200 determines whether data is stored in an unallocated stripe based on the determination result in step S1804. If data is stored in an unallocated stripe, the array controller 200 erases the presence of data on the stripe from the storage state table 203a by updating the storage state table 203c in step S1806. . At this time, for example, the stripe may be initialized by writing “1” in all the blocks constituting the stripe.

  Next, in step S1807, the array controller 200 determines whether or not the processing of S1803 to S1806 has been performed for all unallocated stripes. If the above processing has not been performed for all unallocated stripes, the array controller 200 starts processing for the next unallocated stripe that has not yet been processed in step S1808.

  If it is determined in step S1807 that the processing has been completed for all unallocated stripes, the array controller 200 ends the data erasure processing.

  Next, for example, a case where data stored in an unallocated stripe is erased from the disk array 300 in the state shown in FIG.

  The array controller 200 refers to the conversion table 203b and extracts stripes S0, S1, S5,... As unassigned stripes. First, for the stripe S1, it is determined whether or not data is stored with reference to the storage state table 203a. Here, since the data is stored, the portion related to the stripe S1 in the storage state table 203a is returned to the initial state (see FIG. 4) (see FIG. 19). Subsequently, the storage state table 203a is referred to for the stripe S5. However, since no data is stored, the storage state table 203a is not rewritten (see FIG. 19). Thereafter, the same processing is performed for all unallocated stripes.

  The data erasing process may be performed at a predetermined timing when the array controller 200 is not accessed from the host device 100, for example, when a predetermined time elapses or when the power is turned on. Alternatively, the host apparatus 100 may prepare an instruction for causing the array controller 200 to perform the data erasure process, and the array controller 200 may perform the data erasure process when receiving the data erasure instruction from the host apparatus 100. .

  As described above, according to the present embodiment, in the disk array 300 configured with RAID 4 or RAID 5, the host device 100 among the data written to the disk array 300 as in the first to third embodiments. Delete data that is recognized as unnecessary. As a result, the capacity of the disk array 300 can be used efficiently.

1 is a block diagram showing a configuration of a disk array device according to a first exemplary embodiment of the present invention. 2 is a flowchart showing an operation procedure of the array controller 200 when data is written to the disk array 300 in the disk array apparatus of FIG. It is a flowchart which shows the procedure of the block allocation and parity data generation process of step S202 of FIG. It is a figure which shows each state of the disk array at the time of a format, a data storage state table, and a LBA-disk array block conversion table. It is a figure which shows each state of a disk array, a data storage state table, and an LBA-disk array block conversion table when a data write command for two blocks is issued. It is a figure which shows each state of a disk array, a data storage state table, and an LBA-disk array block conversion table when a data write command for one block is issued. It is a figure which shows each state of a disk array, a data storage state table, and a LBA-disk array block conversion table when a data write command for 4 blocks is issued. It is a figure which shows each state of a disk array, a data storage state table, and an LBA-disk array block conversion table when a data write command for one block is issued. 4 is a flowchart showing an operation procedure of the array controller 200 when reading data from the disk array 300. 10 is a flowchart illustrating a procedure of data restoration processing in step S905 of FIG. 9. It is a flowchart which shows the procedure of the other data restoration process replaced with the data restoration process of FIG. It is a flowchart which shows the procedure of the block allocation and parity data generation process in the disk array apparatus concerning the 2nd Embodiment of this invention. It is a figure which shows each state of the disk array at the time of the data writing in the 2nd Embodiment of this invention, a data storage state table, and an LBA-disk array block conversion table. It is a flowchart which shows the procedure of the block allocation and parity data generation process in the disk array apparatus concerning the 3rd Embodiment of this invention. It is a flowchart which shows the procedure of the data erasure | elimination process in the disk array apparatus based on the 4th Embodiment of this invention. It is a figure which shows each state of the disk array before a data erasing process execution in the 4th Embodiment of this invention, a data storage state table, and a LBA-disk array block conversion table. It is a figure which shows each state of the disk array after a data erasing process execution in the 4th Embodiment of this invention, a data storage state table, and a LBA-disk array block conversion table. It is a flowchart which shows the procedure of the data erasure | elimination process in the disk array apparatus concerning the 5th Embodiment of this invention. It is a figure which shows each state of the disk array after a data erasing process execution in the 5th Embodiment of this invention, a data storage state table, and a LBA-disk array block conversion table.

Explanation of symbols

100 Host device 101 CPU
102 ROM
103 RAM
104 Storage device I / F circuit 200 Array controller 201 MPU
202 ROM
203 RAM
204 Parity arithmetic circuit 205 Host device I / F circuit 206 to 209 HDD I / F circuit 300 Disk array 301 to 304 HDD

Claims (8)

An array type storage device having a storage device array in which a plurality of storage devices are arrayed, and controlling access to the storage device array according to an instruction received from an external device,
Data storage state information storage means for storing data storage state information indicating the presence or absence of data storage for each storage unit in association with each of the redundancy groups including a plurality of storage units configured on the storage device array;
A storage area allocating unit that refers to the data storage state information storage unit and performs storage unit allocation as a storage area for write data and storage unit allocation as a redundant data storage area corresponding to the write data;
Redundant data generating means for generating redundant data corresponding to the write data;
Means for writing the write data and the redundant data to the storage units allocated by the corresponding storage area allocation means,
The storage area allocating means allocates a storage unit included in a redundancy group in which data is not stored in all storage units as a storage area for write data, and writes the write data as an area for writing the redundant data. An array type storage device characterized by assigning another storage unit of a redundancy group including the assigned storage unit. The instruction received from the external device includes a valid instruction for validating the storage area allocation means and the redundant data generation means or an invalid instruction for invalidating the storage area allocation means and the redundant data generation means. ,
When the valid command is received, the storage area allocating unit and the redundant data generating unit are validated. When the invalid command is received, the storage area allocating unit and the redundant data generating unit are invalidated. The array type storage device according to claim 1, wherein: The command received from the external device includes a data integrity critical data write command or a time critical data write command,
When the data integrity critical data write command is received, the storage area allocation unit and the redundant data generation unit are enabled. When the time critical data write command is received, the storage area allocation unit and the redundant data generation unit are activated. 2. The array type storage device according to claim 1, further comprising switching means for invalidation.   Upon receipt of a data write command instructed by the data write destination address from the external device, the storage area allocation unit and the redundant data generation unit are validated based on the data write destination address instructed by the received data write command 2. The array type memory device according to claim 1, further comprising switching means for switching between performing and invalidating. A clock means for measuring time;
Data storage time storage means for storing the time at which data is stored for each storage unit in association with each of the redundant groups including a plurality of storage units configured on the storage device array;
Referring to the data storage time storage means, for a redundancy group in which data is stored in at least one storage unit, the time from the time stored for the storage unit storing the data to the present A predetermined time excess determining means for comparing a time with a predetermined time and determining whether or not the elapsed time exceeds the predetermined time;
The data storage state information of the data storage state information storage unit corresponding to the storage unit in which the data determined that the elapsed time exceeds the predetermined time by the predetermined time excess determination unit is stored in the data. 2. The array type storage device according to claim 1, further comprising data storage state information rewriting means for rewriting data storage state information indicating that the data is not stored.   A storage unit initializing all of the storage units for a redundant group including a storage unit storing data for which the elapsed time has been determined to exceed the predetermined time by the predetermined time excess determining means 6. The array type storage device according to claim 5, further comprising: Address correspondence information storage means for storing information associating a data access destination address included in a data access command received from the external device and a position of a storage unit on the storage device array;
An address non-corresponding redundant group search means for searching for a redundant group that does not include a storage unit corresponding to a data access destination address included in the data access instruction with reference to the address corresponding information storage means;
Rewrite the data storage status information of the data storage information storage means corresponding to each of the storage units of the redundant group found by the address uncorresponding redundant group search means to data storage status information indicating that no data is stored. 2. The array type storage device according to claim 1, further comprising data storage state information rewriting means.   8. The array type storage device according to claim 7, further comprising storage unit initialization means for initializing all of the storage units included in the redundancy group found by said address uncorresponding redundancy group search means.

Images