JP4900203B2 - Storage information arrangement system, storage information arrangement method, and storage information arrangement program - Google Patents

Description

  The present invention relates to a storage data processing load distribution method, and more particularly to a storage information arrangement system, a storage information arrangement method, and storage information for controlling data movement in a system volume including storage such as RAID configured by a plurality of disks. Regarding the placement program.

When I / O processing is concentrated on a logical volume configured on the same RAID group in a disk array device configured from multiple RAID (Redundant Array Inexpensive Disks) groups, the load on this I / O processing (hereinafter “ The processing load is biased to some of the magnetic disks in the device, so the operation of the magnetic disk where the above input / output processing is concentrated becomes a bottleneck, and the processing performance deteriorates in the entire disk array device. there were.
For this reason, when designing the disk array subsystem, consideration must be given to the data arrangement so that input / output processing does not concentrate on the same RAID group. However, when the configuration of the disk array becomes large, a great deal of experience and knowledge is required to realize an optimal data arrangement. Furthermore, when the business requirements change, there is a disadvantage that the system cannot be used effectively with the data arrangement before that.

As a related technology, the processing load between RAID groups is such that I / O processing is not concentrated in one RAID group by automatically relocating data in the disk array device according to the magnetic disk load status between RAID groups. A method of automatically rearranging data according to the situation is disclosed (Patent Document 1).
In this related technology, when one or more logical volumes are prepared as spare volumes in advance for each logical volume group constructed on a RAID group, and a load imbalance is detected between logical volume groups, In this method, data of logical volumes in a group is copied to a spare logical volume in a volume group with a small load, and when the copying is completed, the relocation source volume and the relocation destination volume are exchanged.

JP 2005-209055 A

However, in the related art described in Patent Document 1, since data is sequentially copied at the time of relocation, a large amount of copy processing (input / output processing) for the relocation source logical volume occurs, and simultaneously with normal business processing. Doing so will further increase the load on the logical volume group of the relocation source.
For this reason, even if it is detected that the processing load on the relocation source logical volume group is high, the relocation processing cannot be started until the processing load on the relocation source logical volume group is reduced. It was.
In addition, there is a disadvantage that the logical volume of the relocation destination cannot be used for business processing until the data relocation processing is completed, and therefore, there is a disadvantage that the effect of the load distribution processing by the data relocation does not appear immediately.

[Object of invention]
The present invention provides a storage information arrangement system, a storage information arrangement method, and a storage information arrangement program that improve the inconveniences of the related art, reduce the processing load, and perform a quick and stable operation. Objective.

In order to achieve the above object, in the storage information arrangement system according to the present invention, a storage apparatus having a data storage unit composed of a plurality of storage areas and a cache means for temporarily storing data in the storage areas, and the storage apparatus The storage information placement system includes an input / output device that makes an input / output processing request to the storage device, and the storage device includes a load detection unit that detects an input / output processing load in each storage area and a storage area in advance. When a load deviation equal to or greater than the set value is detected, a storage area with a high load is set as a movement source storage area, and a storage area set in advance corresponding to the movement source storage area is set as a movement destination storage area a placement area setting means for setting, and output control means reads the processed data of the migration source storage to the cache unit prior to the output processing request, the cache hand Rutotomoni a relocation means for storing in the destination storage area processing target data input and output processing with inner processing target data read in the cache unit from the source storage area if stored in the destination area If the storage device has a rearrangement progress monitoring unit that determines that each logical block is allocated if it has been allocated, and if it has not been stored, the block whose input / output control unit has determined that it has been allocated by the rearrangement progress monitoring unit And the predetermined value N / 2 given in advance are compared, and if the number of blocks determined to be arranged is equal to or greater than the predetermined value N / 2, N consecutive logical blocks are selected from the destination storage area. After reading to the cache means, the block determined to be unallocated from the source storage area is read and overwritten in the corresponding area on the cache means, and the number of blocks determined to be allocated is If the value is less than the fixed value N / 2, after reading N consecutive logical blocks from the migration source storage area into the cache means, the blocks determined to have been placed from the migration destination storage area are read and the corresponding on the cache means The area is overwritten .

The storage information arrangement method according to the present invention includes a data storage unit composed of a plurality of storage areas, a cache means for temporarily storing data in the storage areas, and a load detection means for detecting a processing load in each storage area. In the storage device, in a storage information arrangement method for performing input / output processing in response to an input / output processing request for data and performing data rearrangement in the data storage unit, a load monitoring step of monitoring a processing load in each storage area; A processing load deviation detecting step for detecting a bias in processing load relating to input / output processing with respect to the storage area , and when a load deviation greater than a preset value between the storage areas is detected, the storage area having a high load is moved to A storage area setting step for setting a storage area that is set in advance and corresponding to the movement source storage area as a movement destination storage area, A processing target data read no area read ahead cache means the step of the source storage area prior to the output processing request, and reading non-cache read process the processing object data from the source storage area according to the input-output processing request, the cache means A data relocation step for storing the processing target data that has been subjected to input / output processing in the migration destination storage area, and if the processing target data read from the migration source storage area to the cache means is stored in the migration destination area And a rearrangement progress monitoring step for determining for each logical block that it has not been stored if not stored, and the data rearrangement step includes the number of blocks determined to have been placed in the rearrangement progress monitoring step A predetermined value N / 2 given in advance is compared, and if the number of blocks determined to be already arranged is equal to or greater than the specified value N / 2, the destination storage area After the number of consecutive logical blocks is read into the cache means, the block determined to be unallocated from the migration source storage area is read and overwritten in the corresponding area on the cache means, and the number of blocks determined as allocated is specified If the value is less than N / 2, after reading N consecutive logical blocks from the migration source storage area into the cache means, the blocks determined to have been placed from the migration destination storage area are read and the corresponding on the cache means It is characterized by overwriting the area .

Furthermore, the storage information arrangement program according to the present invention includes a data storage unit composed of a plurality of storage areas, a cache means for temporarily storing data in the storage areas, and a load detection means for detecting the processing load of each storage area. A load monitoring function for monitoring the processing load of each storage area in a storage information placement program for performing input / output processing in response to an input / output processing request for data and performing data rearrangement in the data storage unit. , Processing load deviation detection function that detects the bias of the processing load accompanying the input / output processing request and input / output processing, when a load deviation more than a preset value between the storage areas is detected, the storage area with high load A storage area that is set in the movement source storage area and a storage area that is set in advance corresponding to the movement source storage area is set as the movement destination storage area A constant function, a movement source storage area of the processing target data read no area prefetching into the cache means prior to output processing request, read no cache reads processed data from the source storage area according to the input-output processing request Function, a data relocation function for storing processing target data that has undergone input / output processing in the cache means, in the destination storage area, and processing target data read from the source storage area to the cache means in the destination area If the storage device is executed, the relocation progress monitoring function for determining for each logical block that the storage device has been placed is executed , and the data relocation function is used for relocation. The number of blocks determined to be arranged by the progress monitoring function is compared with a predetermined value N / 2 given in advance, and the block decided to be arranged When the number is equal to or greater than the prescribed value N / 2, after reading N consecutive logical blocks from the migration destination storage area into the cache means, the blocks determined to be unallocated from the migration source storage area are read and stored on the cache means If the number of blocks determined to be already overwritten is less than the specified value N / 2, the transfer destination is read after N consecutive logical blocks are read from the transfer source storage area into the cache means. It is characterized in that a block determined to be arranged is read from the storage area and overwritten in the corresponding area on the cache means .

  Since the present invention is configured and functions as described above, according to this, the uneven distribution of the load among the storage areas can be automatically distributed, and the access to the entire storage area can be optimized.

[Embodiment]
In the embodiment of the present invention, as shown in FIG. 1, for example, a processing device (host device) 10 that functions as a client, and a disk array device 100 that performs input / output processing in response to an input / output request sent from the processing device 10. It consists of.
The disk array device 100 includes a physical storage disk 101 as a storage device such as a RAID, and a disk control unit 50 that controls the data stored in the storage disk 101 and the input / output processing.
The disk control unit 50 is connected to the input / output control unit 120 that performs access control of data (volume) stored in the storage disk 101, the input / output processing unit 120, and the host device (input / output device). A cache unit 110 that temporarily stores data, a load monitoring unit 121 that monitors a load deviation between logical volume groups constituting the storage disk 101, and a volume management in which information necessary for this data relocation processing is recorded And a memory 125 that stores a rearrangement target table 131 and a rearrangement progress table 132.
Further, the storage disk 101 is composed of a plurality of (for example, two) RAID groups from a plurality of magnetic disks, and a plurality of logical volume groups are constructed on each RAID group.
In this embodiment, a logical volume group 111 and a logical volume group 112 are configured on two RAID groups.
Each volume group of the logical volume group 111 and the logical volume group 112 includes a plurality of logical volumes, and one or more logical volumes are set in advance as spare volumes for each logical volume group.

  When access is distributed to each logical volume in response to a plurality of I / O requests sent from the host device 10, there is no performance problem. However, if access concentrates on a specific logical volume or logical volume group. The disk device performance cannot be fully exhibited.

  In this embodiment, a logical volume 113 including a logical volume (hereinafter referred to as “transaction volume”) in which the load is concentrated is set in the logical volume group 111, and the logical volume group 112 includes the logical volume 113. The logical volume 114 including the corresponding spare volume is set.

This will be described in detail below.
The processing device 10 is, for example, a host computing device that functions as a client machine. The processing device 10 makes an input / output request to the disk array device 100 and receives input / output processed data sent from the disk array device 100 to perform data processing. Do.

The storage disk 101 has a configuration in which a plurality of magnetic disks, here two RAID groups are constructed, and a plurality of logical volumes are set on the RAID group. Specifically, a logical volume group 111 and a logical volume group 112 are configured on two RAID groups.
The logical volume group 111 and the logical volume group 112 are a set of a plurality of logical volumes, and each logical volume is composed of logical blocks.
Note that a logical block ID is set in advance for each logical block, and the corresponding logical block can be accessed by designating this logical block ID.
Furthermore, as described above, one or more logical volumes are preset as spare volumes for each logical volume group configured on the same RAID group.

  As described above, the memory 125 stores the volume management table 130, the rearrangement target table 131, and the rearrangement progress table 132 in which information necessary for the data rearrangement process is recorded.

Here, the contents of each stored table will be described.
As shown in FIG. 2, there is one volume management table 130 corresponding to each logical volume group, and the capacity and usage (business or spare) of each logical volume constituting the logical volume group are recorded.

  As shown in FIG. 3, the relocation target table 131 is a table that records information of a relocation source volume and a placement destination volume that are being relocated, and is set in the disk array device 100. .

  As shown in FIG. 4, the relocation progress table 132 is a table that records whether or not data relocation has been completed in units of logical blocks. Each pair of relocation source volume and relocation destination volume (FIG. 1). There is one corresponding to (*)). Thereby, the relocation progress table 132 indicates whether the latest data corresponding to each logical block is in the relocation source volume or the relocation destination volume.

  When the load monitoring unit 121 detects an uneven processing load, the input / output control unit 120 causes a business volume in one logical volume group 111 to be in another logical volume group 112 in accordance with an instruction from the load monitoring unit 121. Controls data (volume) relocation processing by copying and moving to a backup volume.

  As shown in FIG. 5, the input / output control unit 120 includes a relocation management unit 140 that controls logical volume relocation processing, a write control unit 142 that performs processing according to instructions from the relocation management unit 140, and read control. It comprises a means 143, a cache control means 141 connected to the cache means 110 for controlling the reading of the processing target data, and a post-processing means 144 for performing the rearrangement processing of the logical block after the rearrangement processing. .

Here, the configuration means of the input / output control unit 120 will be described in detail.
The relocation management unit 140 has an input / output request registration determination function that determines whether or not an access destination logical volume of an input / output request is registered in the relocation target table 131.
Further, when it is determined that the access destination logical volume is registered in the relocation target table 131, the relocation management unit 140 determines that the I / O request type is an output request to the read control unit 143. On the other hand, a processing request notification function is provided for notifying a write request signal to the write control means 142 when the input / output request type is an input request.

Further, the relocation management unit 140 has an information notification function for acquiring information necessary for data relocation processing from the memory and load monitoring unit 121 and notifying the cache control unit 141 of the information.
If the target logical volume is not registered in the relocation target table 131, the input / output control unit 120 performs only normal input / output processing because the target logical volume is not a relocation target.

The cache control unit 141 has a data prefetch acquisition function for reading data that may be accessed from the host device 10 from the target logical volume to the cache unit 110 prior to receiving an input / output processing request.
At this time, the cache control unit 141 has a function of referring to the relocation progress table 132 and checking the relocation status of each logical volume based on the logical volume ID of N logical blocks.

The cache control unit 141 has a data read function and a placement destination data read function for reading the processing target volume (data) from the relocation source and relocation destination volumes into the cache unit 110.
The cache control unit 141 has an input / output processing target volume selection function for selecting a logical volume group to be read / written.

Here, there is only one transaction volume that can be recognized from the host device 10, but during the relocation process, a part of the logical volume is stored on two volumes, the relocation source volume and the relocation destination volume.
For this reason, as will be described later, the cache control unit 141 determines whether the cache unit from the relocation source or the relocation destination volume corresponding to the data based on the relocation state of each logical block (data) in the cache unit 110. 110 has a reading source selection function for determining whether to read data.

The cache control unit 141 has a data read function and a placement destination data read function for reading the processing target volume (data) from the relocation source and relocation destination volumes into the cache unit 110.
As a result, even when a data read request is issued to the relocation source volume during the relocation process, the data that has been relocated can be read from the relocation destination volume to the cache unit.

Here, the cache control unit 141 always reads N consecutive logical blocks (N = predetermined value) that are set in advance in one prefetch process from the logical volume to the cache unit 110.
Further, the cache control unit 141 has a relocation status determination function for checking the relocation status of each logical block by referring to the relocation progress table 132 based on the ID of the N logical blocks based on the key.

  Specifically, among N logical blocks, when a logical block whose rearrangement state is set to “1” (allocated) occupies N / 2 or more, the N consecutive logical blocks are rearranged. A logical block whose relocation status is set to “0” (unallocated) is read from the relocation source volume. At this time, the cache area corresponding to the logical block (logical table) already read in the cache means 110 is overwritten and stored. As a result, the volume (data) read and stored in the cache area 110 is updated to the latest data contents.

On the other hand, when the logical block whose relocation status is “0” occupies N / 2 or more, the N consecutive logical blocks are read from the relocation source volume to the cache 110, and then the relocation status is “1”. The logical block is read from the relocation destination volume and overwritten and stored in the corresponding cache area.
As a result, the continuity of the target volume (data) for I / O processing is assured to the maximum, and the number of I / O processes and accesses to the logical volume can be minimized. The load can be suppressed.

Further, the cache control unit 141 has a relocation status determination function for checking the relocation status of the logical blocks held on the cache unit 110.
In addition, the cache control unit 141 writes the logical block whose relocation status is “0” (unallocated) periodically to be stored in the relocation destination logical volume in the same manner as the updated logical block. Has a return function. Here, after the storage process (write-back) to the relocation destination logical volume is completed, the cache control unit 141 sets the relocation state of the corresponding logical block to “1”.

Further, the cache control unit 141 determines whether or not the relocation status of all logical blocks has become “1”, and when the relocation processing is completed, the cache control unit 141 sets the relocation destination volume. A volume mode conversion setting function is provided for setting a relocation source volume as a spare volume while setting it as a transaction volume.
In addition, the cache control unit 141 has a relocation target initialization function for deleting the registration of the relocated logical volume from the relocation target table 131.

The write control unit 142 includes a WRITE address conversion function that converts the start and end addresses of the WRITE request into logical block IDs (including start and end addresses) of the volume, and a logical block corresponding to the converted address is stored in the cache unit 110. It has a write block read determination function for determining whether or not it exists above.
Here, when the logical block corresponding to the write control unit 142 exists in the cache unit 110, overwriting is performed by executing a write process on the data of the logical block.
On the other hand, when the logical block corresponding to the converted address does not exist on the cache unit 110, the write processing data is stored in the unused cache area, and the cache area (address) is notified to the cache control unit 141.

The read control means 143 has a READ address conversion function for converting the start and end addresses included in the input / output request (READ request) into logical block IDs (including the start address and end address) of the logical volume, and the corresponding logical blocks are It has an R block read determination function for determining whether or not the cache means 110 exists.
Here, if the corresponding logical block exists in the cache unit 110, the data of the logical block is read from the cache unit 110.
If the corresponding logical block does not exist in the cache unit 110, the cache control unit 141 is instructed to specify the logical block ID to be read.

The post-processing unit 144 searches the rearrangement progress table 132 and extracts a logical block whose rearrangement state is “0” when the load on the entire disk array 100 becomes equal to or less than a preset value. It has a logical block extraction function.
Further, the post-processing means 144 has a volume replication arrangement function for sequentially copying the extracted logical block data to the relocation destination volume. At this time, the rearrangement state of the rearrangement progress table 132 is set to “1”.
Further, the post-processing unit 144 has a relocation completion determination function for determining whether or not the relocation status of all the extracted logical blocks has become “1”, and a relocation destination volume when it is determined that relocation has been completed. Is set as a transaction volume, a relocation source volume is set as a spare volume, and a volume reconfiguration function for deleting registration of a relocated volume from the relocation target table 131 is provided.
As a result, data that has never been accessed is moved to the relocation destination volume after the overall load drops to a certain extent, and after all the data has been moved, the relocation destination volume becomes the transaction volume, and the relocation source volume Can be set as a spare volume.

The load monitoring unit 121 is set in advance with a load value detection function for analyzing an input / output processing load in each logical volume and detecting an input / output processing load value, and a detected processing load value (and an updated value). A processing load value registration function for registering in the volume management table 130 is provided.
The load monitoring unit 121 has a processing load deviation detection function for monitoring a processing load value applied to each logical volume group and detecting a processing load deviation between the logical volume groups.
Furthermore, when a processing load deviation exceeding a preset processing load deviation threshold is detected, the load monitoring unit 121 identifies a logical volume (work volume) with a high processing load and supports the logical volume. It has a relocation target setting function that sets the relocated logical volume (transaction volume) as the relocation source volume.

Specifically, the logical volume group 111 with a high processing load is specified, and the volume management table 130 is searched to specify a business volume with the highest processing load, and the volume is set as a relocation source volume.
Also, the volume management table 130 corresponding to the logical volume group 112 having a low load is searched, and it is determined whether there is a spare volume having the same capacity as the set reallocation source volume. Sets the spare volume as the relocation destination volume.

  If there is no spare volume of the same capacity corresponding to the set relocation source volume, the load monitoring unit 121 determines that the relocation source volume data cannot be relocated, and the next A function for selecting a business volume with a high processing load and determining whether a logical volume group with a low processing load has a spare volume corresponding to the business volume is provided. If there is a corresponding spare volume, the spare volume is set as a relocation destination volume.

[Description of Operation of Embodiment]
Next, the operation of the above embodiment will be described.
First, the load monitoring unit 121 monitors the processing load of each storage area (load monitoring process). Detects processing load bias accompanying input / output processing (processing load deviation detection step). Processing target data is read from a storage area with a high processing load (cache reading process), and the processing target data subjected to input / output processing is stored in a storage area with a low processing load by the cache unit (data rearrangement process).

  Here, regarding the load monitoring step, the processing load deviation detection step, the cache reading step, and the data rearrangement step, the execution contents may be programmed and executed by a computer.

  Next, the operation of the load monitoring unit 121 of the disk array device 100 of the above embodiment will be described based on the flowchart of FIG.

  First, the load monitoring unit 121 is always operating, and the input / output processing load of each logical volume in the disk array device 100 is analyzed and recorded in the volume management table 130 (step S101: load monitoring step). When a processing load deviation (deviation exceeding a preset threshold value) is detected between the logical volume groups 111 and 112 (step S102: processing load deviation detection step), the load monitoring unit 121 sets the target to be relocated. Select a logical volume.

Specifically, the volume management table of the logical volume group 111 with the high processing load is searched, and the business volume with the highest processing load is selected as the relocation source volume (step S103).
Next, the volume management table 130 of the logical volume group 112 with a low load is searched to determine whether there is a spare volume having the same capacity as the relocation source volume (step S104).

  If there is a spare volume having the same capacity as the relocation source volume in the logical volume group 112, this spare volume is set as a relocation destination volume (step S105).

  Next, the load monitoring unit 121 registers the relocation source and relocation destination volumes in the relocation target table 131 and creates a new relocation progress table 132 corresponding to the two pair volumes (FIG. 1 (*)). (Step S106).

Next, the “use” field of the volume management table 130 is updated and set to “relocating” (step S107).
On the other hand, if it is determined in step S104 that there is no spare volume of the same capacity, the relocation source volume is disabled, the transaction volume with the next highest load is set as the relocation source volume, and S103 to S104. Perform again.

  Next, the operation of the rearrangement management unit 140 of the input / output control unit 120 will be described based on the flowchart of FIG. The rearrangement management unit 140 operates in response to an input / output request sent from the host device 10.

  First, it is checked whether the access destination volume for the input / output processing request is registered in the relocation target table 131 (step S121). If registered, based on the type of input / output request, in the case of an input (write) request (WRITE request), the write control means 142 (step S122), and in the case of an output request (READ request), read control means The input process or the output process is started by notifying 143 (step S123).

  Next, the operation of the write control means 142 will be described based on the flowchart of FIG. The write control unit 142 operates according to an activation signal notified from the rearrangement management unit 140.

First, the start address and end address of the input request sent are converted into the logical block ID of the logical volume (step S131).
Next, it is checked whether or not the corresponding logical block exists on the cache unit 110 (step S132). If it exists, the corresponding cache area is overwritten (step S133).
If the corresponding logical block does not exist on the cache, data is written (read) in the unused cache area (step S134), and the cache area (address) is notified to the cache control means 141 (step S135).

  Next, the operation of the read control means 143 of the input / output control unit 120 will be described based on the flowchart of FIG. The read control unit 143 operates by being called from the rearrangement management unit 140.

First, the start and end addresses of the read (READ) request are converted into the logical block ID of the volume (step S141). Next, it is checked whether or not the corresponding logical block exists on the cache unit 110 (step S142). If it exists, the data is read from the corresponding cache area (step S143).
If the corresponding logical block does not exist in the cache, the cache control unit 141 is notified of the logical block ID of the necessary data (step S154).

  Next, the operation of the cache control unit 141 will be described based on the flowchart of FIG.

  Prior to the input / output request from the host device 10, the cache control unit 141 reads the business volume into the cache unit 110 based on the result detected by the load monitoring unit 121 (step S151).

  At this time, in one read-ahead process, N (N = predetermined) consecutive logical blocks are always read into the cache, and then the cache control unit 141 relocates progress table based on the IDs of the N logical blocks. 132, the rearrangement state of each logical block is checked (step S152).

Here, when the logical blocks whose relocation status is set to “1” (allocated) occupy N / 2 or more, the N consecutive logical blocks are read from the relocation destination volume to the cache 110 (step S153). ).
Next, the logical block whose relocation status is set to “0” (unallocated) is read from the relocation source volume, and overwritten and stored in the corresponding cache area where the relocation destination volume has been read (step S154).

On the other hand, when the number of logical blocks whose relocation status is “0” occupies N / 2 or more, the N consecutive logical blocks are read from the relocation source volume to the cache unit 110 (step S155).
Next, the logical block whose relocation status is “1” is read from the relocation destination volume and overwritten and stored in the corresponding cache area from which the relocation source volume has been read (step S156).
As a result, the continuity of input / output is assured to the maximum, and the number of input / output processes can be minimized.

  When the transaction volume is read into the cache unit 110 after receiving the input / output request from the host device 10, the cache control unit 141 is based on the logical block ID notified from the write control unit 142 or the read control unit 143. The transaction volume (processing target volume) may be read into the cache unit 110.

  As a result, for the set pair of the reallocation source volume and the reallocation destination volume, the data is prefetched into the cache area 110 and the data on the reallocation source volume is moved to the reallocation destination volume, or the host device The data accessed on the relocation source volume can be moved to the relocation destination volume in response to an input / output request from the tenth.

  Next, the cache control unit 141 checks the relocation status of the logical blocks held on the cache 110 (step S157), and for the logical block whose relocation status is “0” (not allocated). Then, like the updated logical block, it is periodically written back to the relocation destination volume (step S158).

  Next, after the write back is completed, the rearrangement state of the corresponding logical block is set to “1” (step S159). When the relocation status of all logical blocks becomes “1”, the relocation is completed, the relocation destination volume is set as the transaction volume, the relocation source volume is set as the spare volume, and the relocation target table 131 is relocated. The registration of the completed volume is deleted (step S160).

  Finally, the operation of the post-processing means 144 of the input / output control unit 120 will be described based on the flowchart shown in FIG. The post-processing means 144 is executed when the load on the entire disk array becomes a certain level or less (for example, when business is suspended).

  First, the post-processing unit 144 searches the rearrangement progress table 132 and extracts a logical block whose rearrangement state is “0” (step S171). The extracted logical block data is sequentially copied to the relocation destination volume (step S172), and the relocation status of the relocation progress table 132 is set to “1” (step S173). When the relocation status of all logical blocks becomes “1”, the relocation is completed, the relocation destination volume is set as the transaction volume, the relocation source volume is set as the spare volume, and the relocation target table 131 is relocated. The registered volume registration is deleted (step S174).

  In the above embodiment, there are two RAID groups, but even if there are three or more RAID groups, the same can be achieved by selecting the relocation source volume and the relocation destination volume from different RAID groups. Relocation control can be performed by the method.

As described above, in the above embodiment, in the disk array device configured from a plurality of RAID groups, when the input / output processing is concentrated on the logical volume configured on the same RAID group, the processing load is increased in the device. Performance degradation caused by biasing to some magnetic disks can be reduced by automatically performing data relocation according to the magnetic disk load status between RAID groups.
As a result, even in situations where constant data access is expected, such as during online work, automatic data relocation can be performed without affecting the work.

  In addition, there is no load on the read process for extra copy to the relocation source volume, and the latest data in the cache means is rewritten to the relocation destination volume to relocate the data. Data relocation can be realized without imposing a load on the relocation source logical volume group.

  Furthermore, in the present invention, even if all the relocation processing is not completed, all the writing and part of the read processing are redirected to the relocation destination volume. As a result, the load distribution for the transaction volume (relocation source volume) can be distributed immediately after the start of data relocation.

  Data is stored in the cache unit every time a volume is accessed, such as prefetching data or input / output processing from the host device, so the latest data in the cache unit is written back to the relocation destination volume. Relocation can be realized.

  The present invention can be used for automatic data relocation of a database system for online business.

It is a schematic block diagram which shows the whole one Example of the memory | storage information arrangement | positioning system by this invention. It is explanatory drawing which shows an example of the volume management table in the storage information arrangement system disclosed in FIG. It is explanatory drawing which shows an example of the rearrangement object table in the memory | storage information arrangement system disclosed in FIG. It is explanatory drawing which shows an example of the rearrangement progress table in the memory | storage information arrangement system disclosed in FIG. It is a schematic block diagram which shows an example of the input-output control part in the memory | storage information arrangement system disclosed in FIG. 3 is a flowchart showing operation processing steps of a load monitoring unit in the storage information arrangement system disclosed in FIG. 1. It is the flowchart which showed the operation | movement process step of the rearrangement management means in the memory | storage information arrangement system disclosed in FIG. 2 is a flowchart showing operation processing steps of a write control means in the storage information arrangement system disclosed in FIG. 1. 2 is a flowchart showing operation processing steps of a read control unit in the storage information arrangement system disclosed in FIG. 1. 2 is a flowchart showing operation processing steps of a cache control unit in the storage information arrangement system disclosed in FIG. 1. It is the flowchart which showed the operation | movement process step of the post-processing means in the memory | storage information arrangement system disclosed in FIG.

Explanation of symbols

10 Host device 50 Disk control unit 100 Disk array device 101 Storage disk 110 Cache unit 111, 112 Logical volume group 113, 114 Logical volume 120 Input / output control unit 121 Load monitoring unit 125 Memory 130 Volume management table 131 Relocation target table 132 Re Arrangement progress table 140 Relocation monitoring unit 141 Cache control unit 142 Write control unit 143 Read control unit 144 Post processing unit

Claims (7)

A storage device comprising a storage device having a data storage unit composed of a plurality of storage areas and a cache means for temporarily holding data in the storage area, and an input / output device for making an input / output processing request to the storage device An information placement system,
The storage device
Load detection means for detecting the load of input / output processing in each storage area;
When a load deviation greater than a preset value between the storage areas is detected, a storage area with a high load is set as a movement source storage area and a storage area set in advance corresponding to the movement source storage area An arrangement area setting means for setting the movement destination storage area,
Input / output control means for reading processing target data in the migration source storage area into the cache means prior to the input / output processing request;
Relocation means for storing processing target data that has been input / output processed in the cache means in the destination storage area ;
The equipped Rutotomoni,
Relocation progress monitoring for determining for each logical block that the processing target data read from the migration source storage area to the cache means is allocated if it is stored in the migration destination area, and is not allocated otherwise. Means for the storage device,
The input / output control unit compares the number of blocks determined to be allocated by the rearrangement progress monitoring unit with a predetermined value N / 2, and the number of blocks determined to be allocated is determined. If it is equal to or greater than the prescribed value N / 2, after reading N consecutive logical blocks from the migration destination storage area into the cache means, the block determined to be unallocated is read from the migration source storage area. Overwriting the corresponding area on the cache means, and if the number of blocks determined to be allocated is less than the specified value N / 2, N consecutive logical blocks are transferred from the source storage area to the cache storage, characterized in that overwriting from the movement destination storage area after reading the section to the corresponding area on the cache means reads the determined block and the deployed Broadcast placement system. The storage information arrangement system according to claim 1 , wherein
The data storage unit includes a storage area group including a plurality of storage areas,
The storage area setting unit has a function of setting the transfer source and transfer destination storage areas in different storage area groups. In the storage information arrangement system according to claim 2 ,
The relocation unit has a write-back destination selection function for determining whether to store data accessed by the cache unit in the migration source storage area or the migration destination storage area based on the monitoring result. Storage information placement system. In the storage information arrangement system according to claim 2 ,
The input / output control means has a continuously stored data prefetch function for reading a data group continuously stored in the movement source or movement destination storage area prior to the input / output request into the cache means with a preset length; And a cache data update function for reading data from the migration source or migration destination storage area into the cache unit based on the monitoring result of the rearrangement progress and updating the data in the cache unit. Placement system. In the storage information arrangement system according to claim 3 or 4 ,
When data is stored in the mutually corresponding migration source and migration destination storage areas, the input / output control means responds to the input / output process request with respect to the input / output process request to the migration source storage area. A storage information arrangement system comprising a movement destination reading function for reading processing target data from the movement destination storage area into the cache means. A storage apparatus having a data storage unit comprising a plurality of storage areas, a cache means for temporarily holding data in the storage areas, and a load detection means for detecting a processing load in each storage area, In a storage information arrangement method for performing input / output processing in response to an output processing request and performing data rearrangement in the data storage unit,
A load monitoring step for monitoring the processing load in each storage area, a processing load deviation detection step for detecting a bias in processing load for input / output processing with respect to the storage area, and a value greater than or equal to a preset value between the storage areas Storage area setting step of setting a storage area having a high load as a movement source storage area and setting a storage area set in advance corresponding to the movement source storage area as a movement destination storage area when a load deviation is detected An area prefetching step for reading processing target data in the migration source storage area into the cache means prior to the input / output processing request, and reading processing target data from the migration source storage area in response to the input / output processing request a cache read process, the data relocation process of storing the processing object data input and output processing is performed in said cache means to said destination storage area, the mobile A rearrangement progress monitoring step of determining for each logical block that the processing target data read from the storage area is stored in the movement destination area and is determined to be unallocated if not stored. As well as
The data rearrangement step compares the number of blocks determined to have been allocated in the rearrangement progress monitoring step with a predetermined value N / 2 given in advance, and the number of blocks determined to be allocated is determined. If it is equal to or greater than the prescribed value N / 2, after reading N consecutive logical blocks from the migration destination storage area into the cache means, the block determined to be unallocated is read from the migration source storage area. Overwriting the corresponding area on the cache means, and if the number of blocks determined to be allocated is less than the specified value N / 2, N consecutive logical blocks are transferred from the source storage area to the cache serial, characterized in that overwriting from the movement destination storage area after reading the section to the corresponding area on the cache means reads the determined block and the deployed Information placement method. A storage apparatus having a data storage unit composed of a plurality of storage areas, a cache means for temporarily storing data in the storage areas, and a load detection means for detecting a processing load of each storage area, In a storage information arrangement program for performing input / output processing in response to an output processing request and performing data rearrangement in the data storage unit,
A load monitoring function for monitoring the processing load of each storage area, a processing load deviation detection function for detecting a bias in processing load accompanying the input / output processing request and input / output processing, and a value greater than a preset value between the storage areas A storage area setting function for setting a storage area with a high load as a movement source storage area and setting a storage area set in advance corresponding to the movement source storage area as a movement destination storage area when a load deviation is detected And an area prefetch function for reading the processing target data in the migration source storage area into the cache means prior to the input / output processing request, and the processing target data from the migration source storage area in response to the input / output processing request. a cache read function, the data rearrangement function of storing processed data output processing is performed in said cache means to said destination storage area, the migration source storage territory Said storage reading filled-in processed data already arranged if stored in the destination area and is unplaced not stored and relocation progress monitoring function of determining for each logical block in the cache unit from Let the computer provided with the device execute ,
The data rearrangement function compares the number of blocks determined to be allocated by the rearrangement progress monitoring function with a predetermined value N / 2, and the number of blocks determined to be allocated is determined. If it is equal to or greater than the prescribed value N / 2, after reading N consecutive logical blocks from the migration destination storage area into the cache means, the block determined to be unallocated is read from the migration source storage area. Overwriting the corresponding area on the cache means, and if the number of blocks determined to be allocated is less than the specified value N / 2, N consecutive logical blocks are transferred from the source storage area to the cache serial was characterized by overwriting from the movement destination storage area after reading the section to the corresponding area on the cache means reads the determined block and the deployed Information placement program.

Images