Storage device array architecture with copyback cache

• US 5,274,799 A
• Filed: 01/04/1991
• Issued: 12/28/1993
• Est. Priority Date: 01/04/1991
• Status: Expired due to Term

• Current Assignee: EMC Corporation (Dell Technologies Inc.)
• Original Assignee: Array Technology Corp. (Dell Technologies Inc.)
• Inventors: Walker, Mark, Brant, William A., Stallmo, David C., Lui, Albert
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): HUA, LY
• Application Number:
  US07/638,167
• Time in Patent Office: 1,089 Days
• Field of Search: 371/40.1, 371/40.2, 371/51.1, 371/10.1, 371/13, 364/200 MS File, 364/900 MS File, 395/425, 395/575
• US Class Current: 714/6.12
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems
• CPC Class Codes: G06F 11/1435 using file system or storage system metadata
• CPC Class Codes: G06F 11/1441 Resetting or repowering
• CPC Class Codes: G06F 11/1666 where the redundant component is memory or memory area
• CPC Class Codes: G06F 11/20 using active fault-masking, e.g. by switching out faulty elements or by switching in spare elements
• CPC Class Codes: G06F 11/2087 with a common controller
• CPC Class Codes: G06F 11/2089 Redundant storage control functionality
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• CPC Class Codes: G06F 12/0804 with main memory updating G06F12/0806 takes precedence
• CPC Class Codes: G06F 12/0866 for peripheral storage systems, e.g. disk cache
• CPC Class Codes: G06F 2211/1019 Fast writes, i.e. signaling the host that a write is done before data is written to disk
• CPC Class Codes: G06F 2211/1059 Parity-single bit-RAID5, i.e. RAID 5 implementations
• CPC Class Codes: G11B 20/1833 by adding special lists or symbols to the coded information G11B20/1806, G11B20/1866 take precedence

Abstract

A fault-tolerant storage device array using a copyback cache storage unit for temporary storage. When a Write occurs to the RAID system, the data is immediately written to the first available location in the copyback cache storage unit. Upon completion of the Write to the copyback cache storage unit, the host CPU is immediately informed that the Write was successful. Thereafter, further storage unit accesses by the CPU can continue without waiting for an error-correction block update for the data just written. In a first embodiment of the invention, during idle time for relevant storage units of the storage system, an error-correction block is computed for each "pending" data block on the copyback cache storage unit, and the data block and corresponding error-correction block are copied to their proper location in the RAID system. The copyback cache storage unit in effect stores "peak load" Write data and then completes the actual Write operations to the RAID system during relatively quiescent periods of I/O accesses by the CPU. In a second embodiment of the invention, after Write data is logged to the copyback cache storage unit, normal Read-Modify-Write operation by the RAID system controller continues in overlapped fashion with other CPU I/O accesses using Write data in the controller'"'"'s buffer memory. Performance is enhanced because the CPU can continue processing as soon as the simple Write operation to the copyback cache storage unit completes, thus eliminating the delay caused by a normal Read-Modify-Wrote RAID system. In this embodiment, the copyback cache storage unit acts more as a running "log" of Write data.

Claims

1. A fault-tolerant storage device array including:

a. a plurality of failure independent storage units for storing information as stripes of blocks, including at least data blocks and associated error-correction blocks;

b. at least one copyback cache storage unit for temporarily storing data blocks;

c. a storage unit controller, coupled to the plurality of storage units and to the at least one copyback cache storage unit, including control means for;

(1) writing received data blocks initially onto the at least one copyback cache storage unit as pending data blocks;

(2) during idle time of at least some of the plurality of storage units;

(a) reading at least one pending data block from at least one copyback cache storage unit;

(b) accessing the storage units and reading information corresponding to each read pending data block;

(c) generating an associated error-correction block from the read information and each read pending data block;

(d) writing each such read pending data block and associated error-correction block to a corresponding stripe of the idle storage units;

(3) reading requested data blocks from at least one copyback cache storage unit when such requested data blocks have not been written to the plurality of storage units, otherwise from the plurality of storage units.

2. The storage device array of claim 1, wherein the control means acknowledges completion of writing each received data block to the at least one copyback cache storage unit before writing such data block to one of the storage units.

3. The storage device array of claim 1, wherein the information corresponding to each read pending data block includes a corresponding old error-correction block and corresponding old data block read from the corresponding stripe of idle storage units.

4. The storage device array of claim 2, wherein generating a new error-correction block further includes:

a. reading a corresponding old data block from the corresponding stripe of the idle storage units;

b. reading a corresponding old error-correction block from the corresponding stripe of the idle storage units;

c. exclusively-OR'"'"'ing the old data block, the old error correction block, and the read pending data block, thereby generating a new error-correction block.

5. The fault-tolerant storage device array of claim 1, wherein at least one copyback cache storage unit is non-volatile.

6. A method for storing data in a fault-tolerant storage device array comprising a plurality of failure independent storage units for storing information as stripes of blocks, including at least data blocks and associated error-correction blocks, including the steps of:

a. providing at least one copyback cache storage unit for temporarily writing received data blocks;

b. writing received data blocks initially onto the at least one copyback cache storage unit as pending data blocks;

c. during idle time of at least one of the plurality of storage units;

(1) reading at least one pending data block from at least one copyback cache storage unit;

(2) accessing the storage units and reading information corresponding to each read pending data block;

(3) generating an associated error-correction block from the read information and each such read pending data block;

(4) writing each such read pending data block and associated error-correction block to a corresponding stripe of the idle storage units;

d. reading requested data blocks from at least one copyback cache storage unit when such requested data blocks have not been written to the plurality of storage units, otherwise from the plurality of storage units.

7. The method of claim 6, further including the step of acknowledging completion of writing each received data block to the at least one copyback cache storage unit before writing such data block to one of the storage units.

8. The method of claim 6, wherein the step of generating an associated error-correction block from the read information and each read pending data block comprises the steps of:

a. generating a new error-correction block as a function of at least the read pending data block, and a corresponding old error-correction block and corresponding old data block read from the corresponding stripe of the idle storage units.

9. The method of claim 8, wherein the step of generating a new error-correction block comprises the steps of:

a. reading a corresponding old data block from the corresponding stripe of the idle storage units;

b. reading a corresponding old error-correction block from the corresponding stripe of the idle storage units;

c. exclusively-OR'"'"'ing the old data block, the old error-correction block, and the read pending data block, thereby generating a new error-correction block.

10. The method of claim 6, wherein at least one copyback cache storage unit is non-volatile.

11. A fault-tolerant storage device array including:

a. a plurality of failure independent storage units for storing information as stripes of blocks, including at least data blocks and associated error-correction blocks;

b. at least one copyback cache storage unit for temporarily storing data blocks;

c. a storage unit controller, coupled to the plurality of storage units and to the at least one copyback cache storage unit, having a buffer memory and including control means for;

(1) writing received data blocks initially onto the at least one copyback cache storage unit;

(2) temporarily storing received data blocks in the buffer memory as pending data blocks;

(3) during idle time of at least some of the plurality of storage units;

(a) accessing at least one pending data block from the buffer memory;

(b) accessing the storage units and reading information corresponding to each accessed pending data block;

(c) generating an associated error-correction block from the read information and each accessed pending data block;

(d) writing each such accessed pending data block and associated error-correction block to a corresponding stripe of the idle storage units;

(4) reading requested data blocks from the buffer memory when such requested data block have not been written to the plurality of storage units, otherwise from the plurality of storage units.

12. The storage device array of claim 11, wherein the control means substantially immediately acknowledges the completion of writing a received record to the at least one copyback cache storage unit.

13. The storage device array of claim 11, wherein the read information includes a corresponding old error-correction block and corresponding old data block read from the corresponding stripe of the idle storage units.

14. The storage device array of claim 13, wherein generating a new error-correction block further includes:

a. reading a corresponding old data block from the corresponding stripe of the idle storage units;

b. reading a corresponding old error-correction block from the corresponding stripe of the idle storage units;

c. exclusively-OR'"'"'ing the old data block, the old error-correction block, and the accessed pending data block, thereby generating a new error-correction block.

15. The storage device array of claim 11, further including means for reading selected data blocks from the at least one copyback cache storage unit and writing such selected data blocks to the plurality of storage units upon a failure of the storage unit controller to write all corresponding data blocks from the buffer memory to the plurality of storage units.

16. The fault-tolerant storage device array of claim 11, wherein at least one copyback cache storage unit is non-volatile.

17. A method for storing data in a fault-tolerant storage device array comprising a plurality of failure independent storage units for storing information as stripes of blocks, including at least data blocks and associated error-correction blocks, including the steps of:

a. providing buffer memory and at least one copyback cache storage unit for temporarily storing data blocks;

b. writing received data blocks initially onto the at least one copyback cache storage unit;

c. temporarily storing received data blocks in the buffer memory as pending data blocks;

d. during idle time of at least one of the plurality of storage units;

(1) accessing at least one pending data block from the buffer memory;

(2) accessing the storage units and reading the information corresponding to each accessed pending data blocks;

(3) generating an associated error-correction block from the read information and each such accessed pending data block;

(4) writing each such accessed pending data block and associated error-correction block to a corresponding stripe of the idle storage units;

e. reading requested data blocks from the buffer memory when such requested data blocks have not been written to the plurality of storage units, otherwise from the plurality of storage units.

18. The method of claim 17, further including the step of acknowledging completion of writing each received data block to the at least one copyback cache storage unit before writing such data block to one of the storage units.

19. The method of claim 17, wherein the step of generating an associated error-correction block for each accessed pending data block comprises the steps of:

a. generating a new error-correction block as a function of at least the accessed pending data block, and a corresponding old error-correction block and corresponding old data block read from the corresponding stripe of the idle storage units.

20. The method of claim 19, wherein the step of generating a new error-correction block comprises the steps of:

a. reading a corresponding old data block from the corresponding stripe of the idle storage units;

b. reading a corresponding old error-correction block from the corresponding stripe of the idle storage units;

c. exclusively-OR'"'"'ing the old data block, the old error-correction block, and the accessed pending data block, thereby generating a new error-correction block.

21. The method of claim 17, further including the steps of reading selected data blocks from the at least one copyback cache storage unit and writing such selected data blocks to the plurality of storage units upon a failure of the storage unit controller to write all corresponding data blocks from the buffer memory to the plurality of storage units.

22. The method of claim 17, wherein at least one copyback cache storage unit is non-volatile.