Disk control system and method

• US 20020091903A1
• Filed: 09/18/2001
• Published: 07/11/2002
• Est. Priority Date: 01/09/2001
• Status: Active Grant

• Current Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Original Assignee: Kabushiki Kaisha Toshiba (Toshiba Corporation)
• Inventors: Mizuno, Satoshi
• Granted Patent: US 7,216,199 B2
• Time in Patent Office: Days
• US Class Current: 711/154
• CPC Class Codes: G06F 3/0605 by facilitating the interaction with a user or administrator
• CPC Class Codes: G06F 3/0614 Improving the reliability of storage systems
• CPC Class Codes: G06F 3/064 Management of blocks
• CPC Class Codes: G06F 3/0656 Data buffering arrangements
• CPC Class Codes: G06F 3/0689 Disk arrays, e.g. RAID, JBOD

Abstract

In a disk control system having a RAID controller for continuously writing data on a data stripe composed of a plurality disk apparatus, in response to a write request, data blocks are sequentially written on empty areas of a write target data stripe on the plurality of disks in such a manner that at least one data block is written at a time. Further, in response to the write request, logical addresses having address values prior to address translation are written on logical address log areas on the plurality of disks, as logical-address log information. An upper file system is notified that the write has been completed after the data and the logical-address log information have been completely written.

Claims

1. A disk control system that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data to a data stripe as a write area composed of a plurality of disk apparatuses, the system comprising:

means, in response to said write request, for sequentially writing data blocks on empty areas of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time;

means, in response to said write request, for writing said logical addresses from said upper file system on data managing areas provided on the plurality of disks, as logical-address log information; and

means for notifying, in response to the write request from said upper file system, the upper file system that the write has been completed, after said data and said logical-address log information have been completely written.

2. The disk control system according to claim 1, wherein an I/O request is issued for one write request for the plurality of data blocks transferred in said write request by said upper file system, and said plurality of data blocks are simultaneously written to the empty areas of said write target data stripe.

3. The disk control system according to claim 1, wherein said logical-address log information is written on at least one logical-address log area provided in said data managing area.

4. The disk control system according to claim 3, wherein if the data managing area is composed of said plurality of logical-address log areas, one of said logical-address log areas which is not currently used and which was most recently used is used.

5. The disk control system according to claim 3, wherein said logical-address log areas are fixedly provided on said disk.

6. The disk control system according to claim 3, wherein if the system is shut down before said data write process is completed, said logical-address log information recorded in said logical-address log area is used to execute data recovery after system reboot.

7. The disk control system according to claim 1, wherein said data managing area has a stripe ID log area in which a written stripe ID and a stripe ID to be written are recorded.

8. The disk control system according to claim 1, wherein part of said write target data stripe is used as an area in which tag information is written, and logical-address information that has been written on provided is written as said tag information.

9. The disk control system according to claim 8, wherein an address mapping table cache that stores correspondences between logical addresses and physical addresses relating to write data is provided in part of the main memory, and when the data and the tag information have been completely written on said write target data stripe, said address mapping table cache is updated on the basis of said tag information.

10. A disk control system that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the system comprising:

means for writing a plurality of block data corresponding to a plurality of write requests, to a write buffer provided in a main memory;

data write means for responding to said plurality of write requests to simultaneously write all the plurality of data blocks stored in said write buffer on empty areas of an assigned target data stripe of data areas provided on said plurality of disks;

log write means for simultaneously writing said logical addresses from said upper file system corresponding to said plurality of block data, on data areas provided on said plurality of disks, as logical-address log information; and

means for notifying, with respect to the write requests from said upper file system, the upper file system that the writes have been completed, after said data and said logical-address log information have been completely written.

11. The disk control system according to claim 10, wherein the number of said write requests is managed as a variable, and if said variable is larger than a predetermined value, said write requests are retained as pending requests and a batch write process is executed using said data and said log write means, and if said variable is smaller than the predetermined value, a process of writing each of said write requests is executed.

12. The disk control system according to claim 11, wherein whether or not to execute said batch write process is determined when the number of said pending write requests exceeds a predetermined number, or the total of the data sizes of said pending requests equals or exceeds the size of a remaining empty area of said write target data stripe, or a time during which the pending requests have been kept pending exceeds a predetermined value.

13. The disk control system according to claim 10, wherein said logical-address log information is written on at least one logical-address log area provided in said data managing area.

14. The disk control system according to claim 10, wherein if the data managing area is composed of said plurality of logical-address log areas, one of said logical-address log areas which is not currently used and which was most recently used is used.

15. The disk control system according to claim 10, wherein said logical-address log areas are fixedly provided on said disk.

16. The disk control system according to claim 15, wherein if the system is shut down before said data write process is completed, said logical-address log information recorded in said logical-address log area is used to execute data recovery after system reboot.

17. The disk control system according to claim 10, wherein said data managing area has a stripe ID log area in which a written stripe ID and a stripe ID to be written are recorded.

18. The disk control system according to claim 10, wherein part of said write target data stripe is used as an area in which tag information is written, and logical-address information that has been written to said logical-address log area when data for all the areas of said write target data stripe have been provided is written as said tag information.

19. The disk control system according to claim 18, wherein an address mapping table cache that stores correspondences between logical addresses and physical addresses relating to write data is provided in part of the main memory, and when the data and the tag information have been completely written on said write target data stripe, said address mapping table cache is updated on the basis of said tag information.

20. A disk control system that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the system comprising:

means, in response to said write request for sequentially writing data blocks to empty areas of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time;

means, in response to said write request for writing said logical addresses from said upper file system, write data sizes, and checksums of write data to logical-address areas provided on said plurality of disks, as logical-address log information; and

means for notifying, in response to the write request from said upper file system, the upper file system that the write has been completed, after said data and said logical-address log information have been completely written.

21. The disk control system according to claim 20, wherein if the system fails during a write, it is checked whether or not a checksum value being written on said logical-address log area of a stripe being subjected to the write process equals a checksum value determined from the data written on said data area, and if these checksum values are equal, the data are treated as valid, and if these checksum values are not equal, it is determined that the data write has not been completed and the data are discarded.

22. A disk control system that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the system comprising:

means, in response to said write request, for writing said logical addresses from the upper file system, write data sizes, and checksums of data written to a logical-address log area provided on said plurality of disks, as logical-address log information;

means, in response to said write request, for sequentially writing data blocks on empty areas of an assigned target data stripe of data areas provided on the plurality of disks, in such a manner that at least one data block is written at a time; and

means for notifying, in response to the write request from said upper file system, the upper file system that the write has been completed, after said logical-address log information has been completely written in the logical-address log area.

23. The disk control system according to claim 22, wherein a time required to respond to the upper file system is reduced by writing said data blocks on said logical address log area.

25. The disk control system according to claim 24, wherein if said checksums are equal, a list of correspondences between physical and logical addresses is recorded in a TAG area of said data stripe, and the correspondence list in the TAG area is reflected in the address mapping table of the main memory.

24. A disk control system that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the system comprising:

means, in response to said write request, for recording flags indicative of validity or invalidity, stripe ID numbers, and write time stamps for final data in header sections of logical address log areas provided on the plurality of disks, and write time stamps for at least one block data processed by the write request, at least one logical address, and at least one checksum, to entry sections of said logical-address log areas as logical-address log information;

means, in response to said write request, for sequentially writing data blocks on empty areas of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time; and

means, if the system fails during a write, for checking whether or not a checksum value being written on said logical-address log area of a stripe being subjected to the write process and for which a valid flag has been set equals a checksum value determined from the data written on said data area, and for treating the data as valid if these checksum values are equal, while determining that the write has not been completed and discarding the data if these checksum values are unequal.

26. A disk control method that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data to a data stripe as a write area composed of a plurality of disk apparatuses, the method comprising:

responding to said write request to sequentially write data blocks on empty areas of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time;

responding to said write request to write said logical addresses from said upper file system on data managing areas provided on the plurality of disks, as logical-address log information; and

notifying, in response to the write request from said upper file system, the upper file system that the write has been completed, after said data and said logical-address log information have been completely written.

27. A disk control method that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the method comprising:

writing a plurality of block data corresponding to a plurality of write requests, in a write buffer provided in a main memory;

responding to said plurality of write requests to simultaneously write all the plurality of data blocks stored in said write buffer on empty areas of an assigned target data stripe of data areas provided on said plurality of disks;

simultaneously writing all said logical addresses from said upper file system corresponding to said plurality of block data, on data managing areas provided on said plurality of disks, as logical-address log information; and

notifying, with respect to the write requests from said upper file system, the upper file system that the writes have been completed, after said data and said logical-address log information have been completely written.

28. A disk control method that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the method comprising:

responding to said write request to sequentially write data blocks on empty areas of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time;

responding to said write request to write said logical addresses from said upper file system, write data sizes, and checksums of write data on a logical-address log area provided on said plurality of disks, as logical-address log information; and

notifying, in response to the write request from said upper file system, the upper file system that the write has been completed, after said data and said logical-address log information have been completely written.

29. A disk control method that responds to a write request from an upper file system to translate logical addresses into physical ones and then continuously write write-requested data on a data stripe as a write area composed of a plurality of disk apparatuses, the method comprising:

responding to said write request to record flags indicative of validity or invalidity, stripe ID numbers, and write time stamps for final data in header sections of logical address log areas provided on the plurality of disks, and write time stamps for at least one block data processed by the write request, at least one logical address, and at least one checksum, on entry sections of said logical-address log areas as logical-address log information;

responding to said write request to sequentially write data blocks on an empty area of an assigned target data stripe of data areas provided on said plurality of disks, in such a manner that at least one data block is written at a time; and

if the system fails during a write, checking whether or not a checksum value being written on said logical-address log area of a stripe being subjected to the write process and for which a valid flag has been set is equal to a checksum value determined from the data written on said data area, and treating the data as valid if these checksum values are equal, while determining that the write has not been completed and discarding the data if these checksum values are unequal.