JPH11327794A - Data migration method for raid data storage system - Google Patents

Abstract

PROBLEM TO BE SOLVED: To protect data during period of intentional and unintentional stop of system operation. SOLUTION: In this system 100, a migration method to be executed by an array controller 106 is started by selecting the direction of migration, the migration in an descending order is used for every migration to cause reduction of storage capacity and in all other cases, the migration in an ascending order is used. In the migration method, stripes are repeatedly transferred from source structure to destination. Whether an overlapped part exists between the strip to be read and the strip to be written or not is confirmed as a part of each transfer and when the overlapped part is detected, the strip to be overlapped first is transferred to a secured area by the array controller 106. During the migration method, information is held in a non-volatile memory 112 and the migration method is restarted after the intentional or unintentional stop of the system operation by the information by the array controller 106.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

[0001] The present invention relates generally to mass storage devices. In particular, the present invention includes a method for data migration during disk array reconfiguration.

[0002]

2. Description of the Related Art A disk array is a data storage device that combines two or more independent disks into one logical storage device. In general, performance, capacity, and fault tolerance are improved with disk arrays as compared to conventional devices. The disk array uses the RAID concept and
D. originally published in 1987 at the University of California, Berkeley, California by Patterson, Gibson, and Kaz, "Redundant Arrays of Inexpensive Disks,
In this paper, the author describes a disk array type classification method known as RAID level 1-5. Thereafter, the RAID classification method is RAID level 0, etc. Has been extended to include

[0003] For RAID level 0, the data is divided into blocks known as "strips" (a unit of small data) and spread on a system disk as "stripes" (collections of one or more strips). . For example, in a RAID level 0 system consisting of four disks having a 1 kbyte strip size, each 4 kbyte stripe of data is written as four 1 kbyte strips, one on each of the four disks. If the data is 3 kbytes, it is written as three 1 kbyte strips split between three of the four drives. RAID level 0 improves performance and capacity, but does not provide any fault tolerance.

In a RAID level 1 system, each disk is paired with a second disk. The operation of the disk pair is coordinated so that each drive is a mirror image of the pair, and if any disk fails, it is guaranteed that no data will be lost by the remaining disks. In this way, RAID level 1 provides some degree of fault tolerance not found on independent disks.

As in the RAID level 0 system, R
An AID level 3 system stripes data across two or more disks in bits or bytes. RAID level 3 systems have independent parity disks. Each bit stored on the parity disk is the parity of the bit stored at the same location on the remaining data drives. Therefore, the nth bit stored in the parity drive is the nth parity bit stored in each of the remaining drives. Generally, the parity bits are formed by performing an exclusive OR operation. Therefore, the nth bit stored on the parity disk is formed as an exclusive OR of the nth bit stored on the remaining disks. Some level of fault tolerance is provided for RAID level 3 by using a parity drive. In these systems, if any one drive (including the parity drive) fails, the remaining drives can be used to reconstruct or interpolate the data stored in the failed drive.

A RAID level 5 system stripes data using a RAID level 0 strip-by-strip base. RAID level 5 systems also store parity bits to increase fault tolerance. However, unlike RAID level 3, RAID level 5 systems do not have independent parity disks. Instead, at RAID level 5, the parity bits are stored rotating across all disks included in the system. For example, in a RAID level 5 system consisting of three disks, the first data
Parity bits for the stripe can be stored on the first drive. Next, the parity bit for the next stripe of data is stored on the second drive. Next, the parity bit for the third stripe of data is stored on the third drive, and so on. RAID level 5 systems provide the same degree of fault tolerance as RAID level 3 systems. However, RAID level 5 systems often provide higher performance in use.

[0007] RAID level 10 systems are RAID
This is a combination of technologies migrated from level 0 and RAID level 1 systems. These systems are RAID
Data is striped over a series of data as in level 0 systems. These disks are known as primary disks. Further, the RAID 10 system has a mirror disk for each primary disk.
RAID level 10 systems provide fast access for RAID level 0 systems and fault tolerance for RAID level 1 systems.

[0008] Reconfiguration often occurs in the operation of a RAID system. For example, a RAID system can be reconfigured to increase or decrease storage capacity by adding or subtracting disks. RAID systems can also be reconfigured to change RAID levels to change fault tolerance, performance, and storage capacity. Also, the RAID system can be reconfigured to adjust performance by changing other parameters such as the strip size used to perform stripping.

Generally, reconfiguring a RAID system requires transferring data from an old system to a new system. In many cases, this data movement is performed by making a backup copy of the data using a tape drive or other backup drive.
The RAID system is then reconfigured by adding or subtracting drives or changing other parameters. Next, the data is recovered using a tape drive or other backup device.

[0010] Backup and restore methods for transferring data have several disadvantages. The main disadvantage among them is that the associated RAID system is not available until the operation is completed. Often the associated RA
Leaving the ID idle for the required period is impractical and impossible. Also, backup and recovery operations require the operator to perform some manual steps. Each of these steps is prone to error and increases the probability that important data will be destroyed.

[0011] Data migration (reconfiguration), ie, data transfer in changing the configuration of a RAID system, replaces backup and recovery methods for data movement. Basically, data migration copies each data from its pre-reconstruction location to its post-reconstruction location, and can maintain access to the RAID system during data migration. In this way, significant constraints of the data movement backup and recovery method are avoided. Unfortunately, data migration requires a large number of individual disk operations and is therefore often very time consuming. During this time, access to the RAID system can be dramatically slowed, and thus even data migration can be inconvenient and impractical. As a result, there is a need for a system that hasten the reconfiguration of a RAID system.

[0012]

The nature of data migration means that during data migration, it is characterized by a mixed configuration of RAID systems. In a mixed configuration, the migrated data is located at its post-reconstruction location. However, the data to be migrated is still located at its pre-reconstructed position. During this time, the RAID system may not be able to shut down. To make matters worse, a power outage or other unexpected cause of a
The system may be left in an uncertain, partially migrated state. As a result, important data may be lost. Therefore, a need exists for a migration method that protects data during periods of intentional and unintentional system outages.

[0013]

SUMMARY OF THE INVENTION Embodiments of the present invention include a data migration method for reconfiguring a disk array. The preferred environment of the present invention is a reconfigurable RAID system, where the RAID system has multiple data disks, optionally mirror disks,
Has a parity disk. The disk array operates under the control of the array controller. The array controller functions as an interface between the RAID system and any host computer to which it is connected. The array controller has a processor and a memory system. Memory system is non-volatile RAM (NVRAM) that remains enabled during intentional and unintentional shutdowns
Area.

The data storage in a RAID system is characterized by reconfigurable parameters, including the number of data and parity disks contained in the disk array, RAID level, strip size and stripe size. According to the method of the present invention, the RAID system can transition from the (source) configuration before the migration to the destination configuration. The transition from the source configuration to the (destination) configuration after the migration can include a change in RAID level, the number of data and parity disks included in the disk array, and a change in parameters such as strip size and stripe size. . Reconstruction can also be performed to transfer data from one or more disks to one or more replacement disks. Preferably, the method is implemented as a program stored in the memory system of the array controller and executed by the array controller processor.

For the method of the present invention, the array controller first determines whether the reconfiguration will reduce the capacity of the RAID system. For reconfigurations with reduced capacity, migration is configured to work downward starting with the highest logical block address (LBA) contained in the RAID system (descending migration). For all other reconstructions, the migration is configured to work upward starting with the lowest logical block address contained in the RAID system (ascending migration).

When the order of migration is selected,
The array controller calculates the transfer size used to perform the migration. The transfer size is calculated by selecting the number of stripes in the source configuration and the number of stripes in the destination configuration so that the data amount represented by the two numbers is equal. As an example, a RAID level 3 disk system with a 4 Kbyte strip size is RAID level 3 with a 4 Kbyte strip size.
Consider the case of reconfiguring as a 4-disk system. In this case, three stripes in the source configuration represent the same amount of data as two stripes in the destination configuration. Therefore, a transfer size of 24 kbytes is an appropriate transfer size. Although not required, the transfer size is preferably calculated such that the number of source stripes and the number of destination stripes have a least common denominator relationship.

After calculating the transfer size, the array controller repeatedly transfers stripes from the source configuration to the destination configuration. Each transfer contains several stripes equal to the calculated transfer size. As part of each transfer, the array controller determines if there is an overlap between the stripe read from the source configuration and the stripe written to the destination configuration. In particular, the array controller checks whether any stripes to be transferred are read and written. If there are no overlaps, the array controller transfers each stripe from its location in the source configuration to its corresponding location in the destination configuration.

In another case (ie, if there is an overlap), the array controller first copies each overlapping strip to a reserved area in the disk array.
After performing this backup operation, the array controller copies each strip to its corresponding location in the destination configuration. When the migration operation is interrupted, it is possible to safely recover from the area where the overlapping strips are secured. Next, the array controller repeats the sequence of steps until all stripes included in the source configuration have been transferred to the destination configuration.

The array controller updates the NVRAM during each step of the migration method. In particular, the array controller maintains a migration pointer in NVRAM. The migration pointer indicates the last logical block address that has been migrated from the source configuration to the destination configuration. Also, NVR
The AM also contains data describing the identity and location of the duplicate strip being transferred to the reserved area. The information maintained by the array controller in NVRAM allows the migration method to be restarted upon intentional or unintentional shutdown of the system.

[0020]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Reference will now be made in detail to preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings. Like reference numerals throughout the drawings denote the same or similar parts.

Environment As shown in FIG. 1, a typical environment of the present invention includes a reconfigurable RAID system 100. RAID
The system 100 has a disk array 102 composed of one or more disks, and a disk 104 therein.
a to 104e are shown. In the exemplary environment of FIG. 1, disks 104a-104d are shown as data disks and 104e is shown as a parity disk. The disk 104 can be configured as a mirror disk. In practice, a mix of data, mirror, and parity disks is a RAID system 10
0 between different embodiments, RAID system 1
00 specific RAID configuration. Disc 1
04 represents a wide variety of commercially available or suitable types, including SCSI, IDE and other disk drive types.

The disk array 102 operates under the control of an array controller 106. Array controller 10
6 functions as an interface between the RAID system 100 and any host system to which it is connected.
The array controller 106 includes a processor 108, a memory 110, and a nonvolatile RAM (NVRAM) 112. The processor 108 executes a program contained in the memory 110 to control the operation of the disk array 102. The processor 108 uses the NVRAM 112 as a permanent data storage area. Data stored in NVRAM 112 by processor 108 remains valid during intentional and unintentional shutdown of RAID system 100.

The array controller 106 maintains one or more reserved storage areas 114 in the RAID system 100. In the case of FIG. 1, the storage area 114 is reserved on each disk 104 (the storage area 114a reserved for the disk 104a, the storage area 114b reserved for the disk 104b, etc.). RAID system 1
More or less storage area 11 within 00
4 may be practical. The secured area 114 functions as a permanent data storage area. Thus, data stored by processor 108 in reserved area 114 remains valid during intentional and unintentional shutdown of RAID system 100.

Embodiments of the present invention include a method and apparatus for reconfiguring the RAID system 100 for fault tolerance. With this method and apparatus, the RAID system 100
Can be reconfigured to change the number and type of disks, RAID level, strip size and stripe size. The RAID system 100 is kept online during the reconfiguration and the RAID system 10
Reconfiguration can subsequently be resumed during zero intentional and unintentional shutdowns.

To provide fault tolerance reconfiguration, the array controller 106
Maintain some data structures in 00. As shown in FIG. 2, these data structures include a primary description 200, a secondary description 202, and a migration pointer 204. Each of these data structures is maintained during the work of fault tolerance. In the case of the migration pointer 204, the fault tolerance is NVRAM
Preferably, this is achieved by recording the migration pointer 204 in 112. Primary descriptions 200 and 2
In the case of the next description 202, the fault tolerance is preferably achieved by storing the primary description 200 and the secondary description 202 in the secured storage area 114. Migration pointer 204, primary descriptions 200 and 2
Storing the next description 202 in the NVRAM 112 and the secured storage area 114 is a typical example. In other embodiments, these elements can be selected to be stored in other fault-tolerant manners.

The primary description 200 is the RAID system 100
Describe the operation parameters of During normal operation, the array controller 106 uses the data contained in the primary description 200 to translate instructions received by the RAID system 100 into corresponding operations directed to the disk array 102. In particular, array controller 106 receives requests to read and write data contained in disk array 102. These requests are directed to one or more logical block addresses (LBAs). Array controller 106 translates each received request into one or more requests for disks 104 included in disk array 102. Each of these requests is directed to one or more drive logical block addresses (DLBAs). The array controller 106 performs conversion from the LBA to the corresponding DLBA using the information included in the primary description 200.

Similar to the primary description 200, the secondary description 202
Functions as a description of the RAID system 100. However, unlike the primary description 200, the secondary description is RA
Used only during reconfiguration of the ID system 100. In this specification, the RAID system 10 before the migration
A configuration of 0 is called a source configuration. Similarly, the configuration of the RAID system 100 after migration is called a destination configuration. During migration, the data contained in the primary description 200 describes the source configuration. On the other hand, the data included in the secondary description 202 describes the destination configuration.

During reconfiguration, the array processor 106 uses the migration pointer 204 to distinguish between LBAs contained in the source configuration and LBAs contained in the destination configuration. In particular, as part of the reconfiguration process, array controller 106 performs a migration operation to repeatedly relocate data associated with each LBA from its location in the source configuration to its location in the destination configuration. Since this process is not a small process, there is a period when the RAID system 100 is characterized by a mixed configuration. In a mixed configuration, some LBAs are properly mapped to DLBAs in the destination configuration. Other LBAs are properly mapped to DLBAs contained in the source configuration. Array controller 1
06 maintains the value of the next LBA to be migrated in the migration pointer 204. In this manner, the array controller 106 is configured to include the LBA included in the source configuration and the LBA included in the destination configuration.
And can be distinguished.

FIG. 3 is a flowchart illustrating a method of reconstructing the fault tolerance of the RAID system 100.
Shown as 0. Method 300 symbolically begins with a start placeholder 302. In the start placeholder 302, the primary description 200 contains information describing the source configuration. Step 3 in the start placeholder 302
04 follows. In step 304, the array controller 106 checks whether the overall capacity of the RAID system 100 is reduced as a result of the reconfiguration. Disk 104
Is reduced from the RAID system 100, a reduction in capacity occurs. Reconfigure the RAID system 100 between RAID levels without a corresponding increase in the number of disks 104 (change from RAID level 0 to any other RAID level, or from any RAID level to RAID
(Change to level 1) also causes a capacity reduction. The array controller 106 refers to logical equivalents of the following types of tables to determine whether a capacity reduction occurs.

[0030]

[Table 1]

In the preceding table, the horizontal axis represents the original configuration. The vertical axis represents the destination configuration. The label indicates the RAID level and the drive count, and is “0”.
"-6" means RAID0-6 drive, and "3 / 5-
"4" indicates a RAID3 or RAID5-4 level. "=" Indicates no capacity change, "+" indicates capacity increase, and "-" indicates capacity decrease.

If the array controller 106 determines that the reconfiguration results in a reduction in the capacity of the RAID system 100, execution of the method 300 continues at step 306. At step 306, the array controller reconfigures the RAID system 100 to reflect the capacity reduction required for migration from the source configuration 500 to the destination configuration 600. RA
Reducing the capacity of the ID system 100 allows the array controller 106 to
Request to store more data in the RAID system 100 than the capacity of the RAID system 100 is prevented. Method 300
This precautionary measure is necessary because the system can be used when the RAID system 100 operates.

In step 308, the array controller 106 initializes the secondary description 202 and describes the destination configuration. In step 310, the array controller 106 determines that the migration pointer 2
04 is initialized to the value before migration. The pre-migration value used in step 310 depends on the type of migration used. In particular, to perform the migration, the array controller 106 repeatedly sequences each LBA included in the RAID system 100. At each LBA, the array controller 106 moves all data addressed by the LBA from its location in the source configuration to its corresponding location in the destination configuration. Migrations can occur in any predetermined order. In particular, the migration can start with the first LBA included in the RAID system 100 and complete with the last LBA (ascending migration). Further, the migration can be started with the last LBA included in the RAID system 100 and completed with the first LBA (downward migration). In step 310, the migration pointer 204 is initialized to point to the first LBA to be migrated. In the case of ascending migration, the migration pointer 204 is initialized to point to the first LBA contained in the source configuration. In the case of descending migration, the migration pointer 204 is initialized to point to the last LBA included in the destination configuration.

In step 312, the data contained in RAID system 100 is migrated from the layout required for the source configuration to the layout required for the destination configuration. As described above, the array controller 106 sequences each LBA included in the RAID system 100. At each LBA, the array controller 106 moves all data addressed by the LBA from its location in the source configuration to its corresponding location in the destination configuration. Array controller 1
06 increments (for ascending migrations) or decrements (for descending migrations) during each iteration of the migration operation. Therefore, the migration pointer 204 always points to the next LBA to be migrated.

The array controller 106 determines in step 31
Upon completing the second migration operation, the method 300
Reaches step 314. In step 314,
The primary description 200 is updated to reflect that the migration of the array controller 106 has been completed.
To perform this update, the array controller 106
Copies the information contained in the secondary description 202 to the primary description 200. Thus, the primary description 200 has the RAI
The D system 100 is updated so as to reflect the operation parameters after the reconfiguration.

In step 316, the array controller 106 determines whether the overall capacity of the RAID system 100 increases as a result of the reconfiguration. Disk 104
Is added to the RAID system 100,
A capacity increase occurs. RAID system 100 is RAID
If reconfigured between levels (eg, any RAID
Change from level to RAID level 0, or RAI
A change from D level 1 to any RAID level) also causes an increase in capacity. Generally, the array controller 106
Can check this by referring to an appropriate table stored in the memory 110 or the NVRAM 112.

If the array controller 106 determines that the reconfiguration will cause an increase in the capacity of the RAID system 100, execution of the method 300 continues at step 318. In step 318, the array controller reconfigures the RAID system 100 to reflect the increased capacity created by the reconfiguration. The array controller 106 effectively limits access to any additional capacity until the reconfiguration is complete. In this way, the array controller 106 avoids processing requests to store data that is accommodated in the destination configuration but exceeds the partial reconfiguration capacity of the RAID system 100. This precaution is necessary because the method 300 can be used during operation of the RAID system 100. Execution of method 300 stops placeholder 3
Ends with 20. Also, the array controller 106 can be configured to dynamically redirect requests directed to LBAs added by reconfiguration.

RAID system 100 is maintained accessible during the execution of method 300. As a result, the array controller may receive a request to access data stored in the RAID system 100 during the execution of the method 300. For each request received, the array controller compares each LBA included in the request with the migration pointer 204. If the comparison indicates that the LBA has not yet been migrated, the array controller 106 uses the information contained in the primary description 200 to map the LBA to its corresponding DLBA. If the comparison indicates that the LBA has already been migrated, the array controller 106
The LBA is mapped to its corresponding DLBA using the information contained in the following description 202. Thus, RA
The ID system 100 is maintained online, accessible during the execution of the method 300.

The persistence of the migration pointer 204, the primary description 200 and the secondary description in the persistent storage means that the method 300 is fault tolerant. Thus, the method 300 can be restarted after a power outage or other system outage. Optimized Migration FIG. 4 shows a flowchart of an optimized migration method as method 400. Method 400 is RAID
It is used during the reconfiguration of the system 100. In particular, the method 400 can be used to perform the migration of step 316 of the method 300. Method 400 symbolically begins with a start placeholder 402. At the start placeholder 402, it can be assumed that migration is performed from the source configuration 500 of FIG. 5 to the destination configuration 600 of FIG.

The source configuration 500 is the RAID system 10
0 corresponds to a RAID 0 configuration. Therefore, the source configuration 500 has two columns 502
a, 502b. Column 502a represents a strip stored on first disk 104. Column 5
02b represents a strip stored on the second disk 104. The destination configuration 600 corresponds to a three-disk, RAID0 configuration of the RAID system 100. Therefore, the destination configuration 60
0 includes three columns 602a to 602c. Column 602a represents a strip stored on first disk 104. Column 602b is the second disk 104
Represents the strip stored above. Column 602c represents a strip stored on third disk 104.

The start placeholder 402 is followed by step 404. In step 404, the array controller 106 calculates the transfer size N. Transfer size N
Is calculated by selecting the number of stripes in the source configuration 500 and the number of stripes in the destination configuration 600 such that the amount of data represented by the two numbers is equal. As an example, consider the case where each strip included in source configuration 500 and destination configuration 600 contains 2 kbytes of data. In this case, the three stripes in the source configuration 500 represent the same amount of data as the two stripes in the destination configuration 600. Therefore, the transfer size of 12 kbytes is an appropriate value for the transfer size N. Although not required, preferably
The transfer size N is calculated to correspond to the total number of source stripes and the total number of destination stripes. Here, it is assumed that the 12 Kbyte value has been selected by the array controller 106 as a value for N.

Next, the array controller 106 enters a loop including steps 406 to 416. During each trip through this loop, the array controller 106 transfers N bytes in size from the source configuration 500 to the destination configuration 600. At step 406, array controller 106 selects a source range. For ascending migration, the source range begins with the value contained within migration pointer 204. The source range ends with an offset equal to the value contained in the migration pointer 204 plus the width of the transfer size N. For descending migration, the source range ends with the value contained in the migration pointer 204. Source range is migration pointer 204
Starting with an offset equal to the value contained in the minus the width of the transfer size N. FIG. 5 shows that the source range constructed during the first iteration of step 414 corresponds to a strip numbered 1-6. Array controller 106 blocks incoming requests directed to LBAs that fall within the source range. These requests are queued and queued until the strips within the source range are transferred to the destination configuration 600.

In step 408, array controller 106 maps the source range selected in step 406 into a corresponding destination range. The destination range is a one-to-one mapping of the strips included in the source range to corresponding locations in destination configuration 600. This mapping is performed by the array controller 106 using the information contained in the primary description 200 and the secondary description 202. FIG.
In, the destination range constructed during the first iteration of step 408 corresponds to a strip numbered 1-6.

In step 410, the array controller 106 determines whether there is an overlap between the source range and the destination range. In particular, the array controller 106 determines whether any of the strips included in the source area are completely or partially included in the destination area. This evaluation can be performed using various methods. One such method is to define start and end DLBA offsets for the source and destination ranges. The starting DLBA offset of the source range is the DLBA of the first strip in the source range. The ending DLBA offset of the source range is the DLBA of the last strip in the source range. Similarly, the start and end DLBA offsets of the destination range are the DLBAs of the first and last strips, respectively, in the destination range. Using this method, the array controller 106 then starts the source range start DLB
The A offset and the ending DLBA offset are compared with the start and end DLBA offsets of the destination range. FIG. 7 schematically illustrates strips 1 through 5 of the source configuration 500, showing that these strips overlap between the source and destination ranges.

If the array controller 106 determines that there is an overlap between the source and destination ranges, the method 400 continues at step 412. At step 412, the array controller 106 copies all duplicate strips included in the source range to the reserved storage area 114 included in the RAID system 100. FIG. 8 shows strips 1-4 of source configuration 500 after being transferred to a reserved area in RAID system 100. After copying the duplicate strip to the reserved area, the array controller 106 updates the NVRAM to indicate that the duplicate strip has been copied to the reserved area 114.

At step 414, array controller 106 copies the strips contained in the source range to the corresponding locations in the destination range. FIG. 9 shows the source configuration 500 after the strips contained within the source range have been transferred to the destination range.
During step 414, the array controller 106 also updates the migration pointer 204. For ascending migration, the migration pointer 204 is incremented by the width of the transfer size N. For descending migration, the migration pointer 204 is decremented by the width of the transfer size N.

In step 416, the array controller 106 checks whether there are any more strips to be transferred. For ascending migration, the array controller 106 sets the migration pointer 204 to the highest LB included in the source configuration 500.
A and maximum L included in destination configuration 600
This confirmation is made in comparison with the smaller one of the BAs. For descending migration, the array controller 10
No. 6 makes this check by comparing the migration pointer 204 with zero. In the positive case (i.e., the migration to destination configuration 600 has not yet been completed), the loop of steps 406-416 is repeated one more time and execution of method 400 continues.

In the negative case (ie, migration to destination configuration 600 has been completed), execution of method 400 ends at stop marker 418. During the execution of method 400, the RAID system is maintained accessible. Since the data is only moved to the reserved area when needed (ie, when a duplicate migration operation is detected), the time required to perform method 400 is minimized. Method 400 is also fault tolerant. In particular, following an interruption, the array controller uses the copy stored in the reserved area 114 to complete any incomplete transfer of the duplicate strip. This prevents potential alterations that may occur if duplicate strips are copied from their original location. Thus, after a power outage or other system shutdown, the method 400 can be restarted.

FIG. 10 illustrates a method 1000 for reducing the occurrence of duplicate migration operations during reconfiguration of the RAID system 100 using a flowchart as method 1000. Method 10
00 can be used in combination with the migration method 400 of FIG. 4 and the reconfiguration method of FIG. Symbolically, the method 1000 starts with placeholder 1
It starts with 002.

The start placeholder 1002 is followed by step 1004. In step 1004, the array controller 106 determines that the RAID system 1
Check whether the total capacity of 00 decreases. As described above, this check can be made with reference to an appropriate table. In the positive case (i.e., the reconfiguration of the RAID system 100 reduces the capacity), the execution of the method 1000 continues at step 1006.
In step 1006, the array controller 106
Allows descending migration.

In the negative case (ie, no capacity reduction due to reconfiguration of RAID system 100), execution of method 1000 continues at step 1008. In step 1008, the array controller 1
06 enables ascending migration. In either case, execution of method 1000 ends at stop placeholder 1010.

In the method 1000, the array controller 1
In step 06, it is effective to select the migration order based on the storage capacity of the source and destination configurations. In the case where the storage capacity is reduced by the reconfiguration from the source configuration to the destination configuration, the array controller 106 selects the descending migration.
In all other cases, the array controller 106 selects descending migration. By selecting a migration order in this manner, the array controller 106 minimizes duplicate migration operations. In this way, the migration speed is increased.

The advantages of the invention will be set forth in part in the description that follows, and will be understood in part by those skilled in the art from the description. The advantages of the invention will be realized and attained by means of the elements and combinations particularly pointed out in the appended claims and equivalents thereof.

[Brief description of the drawings]

FIG. 1 is a block diagram of a RAID system shown as a representative environment of the present invention.

FIG. 2 is a block diagram showing a data structure used in an embodiment of the present invention.

FIG. 3 is a flowchart showing steps related to a reconstruction method according to an embodiment of the present invention.

FIG. 4 is a flowchart showing steps related to a migration method according to the embodiment of the present invention.

FIG. 5 is a block diagram of a source configuration for the RAID system of FIG. 1;

FIG. 6 is a block diagram of a destination configuration for the RAID system in FIG. 1;

FIG. 7 with a group of overlapping strips shown in dashed lines
FIG. 2 is a block diagram showing a source configuration of FIG.

FIG. 8 is a block diagram showing the source configuration of FIG. 3 after moving to the area where the overlapping strip of FIG. 6 is secured.

FIG. 9 is a block diagram showing the source configuration of FIG. 3 after partial migration to the destination configuration of FIG. 4;

FIG. 10 is a flow diagram illustrating steps associated with an embodiment of a method for reducing duplicate migration operations.

[Description of Signs] 100 RAID system 102 Disk array 104a, 104b, 104c, 104d, 114a,
114b, 114c, 114d Data disk 104e, 114e Parity disk 106 Array controller 108 Processor 110 Memory 112 Non-volatile RAM (NVRAM)

──────────────────────────────────────────────────の Continued on the front page (51) Int.Cl. ⁶ Identification code FI G11B 19/02 501 G11B 19/02 501F

Claims (21)

[Claims] 1. A method for reconfiguring a disk array, comprising: providing a description of a configuration before reconfiguration and a description of a configuration after reconfiguration; and reducing the capacity of the disk array as a result of the reconfiguration. A step of confirming whether or not the disk array is to be reconfigured, a step of configuring the disk array so as to deny access exceeding the capacity of the reconfigured configuration if the capacity is reduced as a result of the reconfiguration, and Repeatedly moving data blocks to the reconstructed configuration. 2. The method of claim 12, further comprising: receiving a request to access the disk array; verifying whether a requested data block has been moved; Accessing the requested data block using the pre-reconstruction description if the block has not been moved, and reconstructing if the requested data block has already been moved Accessing the requested data block using the description of the later configuration. 3. The method according to claim 1, further comprising: checking whether the capacity of the disk array increases as a result of the reconfiguration; and reconfiguring if the capacity increases as a result of the reconfiguration. Configuring the disk array to accept access to the capacity of the later configuration. 4. The method of claim 1, further comprising:
A method comprising replacing a description of a configuration before reconfiguration with a description of a configuration after reconfiguration. 5. The method of claim 1, further comprising:
Determining the transfer size, wherein the transfer size corresponds to the M stripes in the configuration before the reconfiguration and the N stripes in the configuration after the reconfiguration, and the M stripes in the configuration before the reconfiguration are different from those in the configuration after the reconfiguration. A method selected to have as many bytes as N stripes. 6. The method of claim 1, further comprising: selecting a descending order migration if the capacity of the disk array decreases as a result of the reconfiguration; and increasing the capacity of the disk array as a result of the reconfiguration. If so, selecting ascending migration. 7. The method of claim 1, further comprising: detecting whether an iteration of the migration involves overlapping data blocks in the configuration before and after the reconfiguration. A method comprising: storing data blocks in a reserved area of a disk array; and storing duplicate data blocks in a reconstructed configuration. 8. A computer program product comprising a computer usable medium containing computer readable code, such that the computer readable program code device implements a method for reconfiguring a disk array. The method comprising: providing a description of the configuration before reconfiguration and a description of the configuration after reconfiguration; determining whether the capacity of the disk array is reduced as a result of the reconfiguration; If the capacity decreases as a result of the configuration, a step of configuring the disk array so as to reject accesses exceeding the capacity of the configuration after the reconfiguration, and repeatedly moving data blocks from the configuration before the reconfiguration to the configuration after the reconfiguration Causing the computer program product to include: 9. The computer program product of claim 8, wherein the method further comprises: receiving a request to access the disk array; and determining whether a requested data block has been moved. Accessing the requested data block using the description of the configuration before reconfiguration, if the requested data block has not been moved, and the requested data block has already been moved. Accessing the requested data block using the description of the reconstructed configuration, if applicable. 10. The computer program product of claim 8, wherein the method further comprises: determining whether a capacity of the disk array increases as a result of the reconstruction; and increasing the capacity as a result of the reconstruction. In some cases, configuring the disk array to accept access up to the capacity of the reconfigured configuration. 11. The computer program product of claim 8, wherein the method further comprises the step of replacing a description of the configuration before reconfiguration with a description of the configuration after reconfiguration. 12. The computer program of claim 8, wherein the method further comprises the step of determining a transfer size, wherein the transfer size is M in the configuration before reconfiguration.
A computer program product corresponding to the stripes and N stripes in the reconfigured configuration, wherein the M stripes in the configuration before reconfiguration are selected to have the same number of bytes as the N stripes in the configuration after reconfiguration. 13. The computer program product of claim 8, wherein the method further comprises: selecting a descending migration if the capacity of the disk array is reduced as a result of the reconfiguration; and the result of the reconfiguration. Selecting an ascending migration if the capacity of the disk array increases. 14. The computer program product of claim 8, wherein the method further comprises detecting whether the iteration of the migration involves overlapping data blocks in the configuration before and after the reconfiguration. Computer program product comprising: storing duplicate data blocks in a reserved area of a disk array; and storing the duplicate data blocks in a reconstructed configuration. 15. A system for reconfiguring a disk array, comprising: means for causing a processor to provide a description of a configuration before reconfiguration and a description of a configuration after reconfiguration; Means configured to determine if the capacity of the disk array is reduced as a result of the reconfiguration, and if the capacity is reduced as a result of the reconfiguration, the processor is refused access beyond the capacity of the reconfigured configuration. And a means configured to cause the processor to repeatedly move data blocks from the configuration before the reconfiguration to the configuration after the reconfiguration. 16. The system according to claim 15, wherein:
A means configured to cause the processor to receive the request to access the disk array; a means configured to cause the processor to determine whether the requested data block has been moved; and a requested data block. Has been moved, means for causing the processor to access the requested data block using the description of the configuration before reconfiguration, and the requested data block has already been moved. And, if so, means configured to cause the processor to access the requested data block using the description of the reconfigured configuration. 17. The system according to claim 15, wherein
In addition, a means for causing the processor to check whether the capacity of the disk array increases as a result of the reconfiguration, and in the case where the capacity increases as a result of the reconfiguration, the processor is allowed to accept access up to the capacity of the reconfigured configuration. Means configured to configure a disk array. 18. The system according to claim 15, wherein
A system further comprising means configured to cause the processor to replace the configuration description before reconfiguration with the configuration description after reconfiguration. 19. The system according to claim 15, wherein
The apparatus further includes means for causing the processor to determine a transfer size, wherein the transfer size corresponds to the M stripes in the configuration before reconfiguration and the N stripes in the configuration after reconfiguration, and the M stripes in the configuration before reconfiguration are reconfigured. A system selected to have as many bytes as N stripes in later configurations. 20. The system according to claim 15, wherein
Further, means configured to cause the processor to select a descending migration if the disk array capacity is reduced as a result of the reconfiguration, and an ascending order to the processor if the disk array capacity is reduced as a result of the reconfiguration. Means configured to cause a migration to be selected. 21. The system according to claim 15, wherein
Further, means configured to cause the processor to detect whether the repetition of the migration involves overlapping data blocks in the configuration before and after the reconfiguration; and A system comprising: means configured to store in a reserved area; and means configured to cause a processor to store overlapping data blocks in a reconstructed configuration.