Method and apparatus for preserving data integrity in a multiple disk raid organized storage system

• US 5,504,858 A
• Filed: 02/08/1995
• Issued: 04/02/1996
• Est. Priority Date: 06/29/1993
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Digital Equipment Corporation (HP Inc.)
• Inventors: Malan, Steven J., Wells, Richard B., Rivera, Peter, Snyder, Sid, Ellis, Robert A., Lubbers, Clark E., Thiel, David W.
• Primary Examiner(s): Beausoliel, Jr., Robert W.
• Assistant Examiner(s): Palys, Joseph E.
• Application Number:
  US08/385,227
• Time in Patent Office: 419 Days
• Field of Search: 395/185.05, 395/182.04, 395/182.03, 395/441
• US Class Current: 714/6.24
• CPC Class Codes: G06F 11/1076 Parity data used in redundant arrays of independent storages, e.g. in RAID systems

Abstract

A data storage system using a RAID array and a logging process and apparatus that identifies a parity block that may not contain the proper parity for its associated data blocks and which prevents such inconsistent parity information from being used in the regeneration of unavailable data. A small fraction of the blocks of each disk are dedicated to storing parity metadata bits. The parity metadata is associated with the parity blocks and identifies whether or not each parity block contains the proper parity information for its associated data blocks or may contain invalid information. The data integrity of the RAID array is preserved by preventing the generation of undetected corrupt data.

Claims

1. A method of operating an array of storage devices, comprising the steps of:

receiving a write request from a processor coupled to said array of storage devices, said write request containing data to be stored on said array of storage devices;

identifying, in response to said write request, a metadata unit data structure (MDU) stored on said array of storage devices to receive said data, said MDU containing a parity block and a set of data blocks, both blocks comprising of a plurality of bits, the bits of said parity block reflecting the parity of a set of corresponding bits from said data blocks when said parity block and said data blocks are consistent with respect to each other, said MDU also including parity metadata (PMD) providing information about whether said parity block and said data blocks are consistent;

setting said PMD to a value indicating that said parity block and said data blocks may be inconsistent;

writing said data to one of said data blocks after said PMD has been set to said inconsistency-indicating value;

updating said parity block to reflect the new parity of said data blocks upon the writing of said data to said one data block, said parity block being updated after said PMD has been set to said inconsistency-indicating value and either before or after said data has been written to said one data block;

determining whether the writing of said data and the updating of said parity block have both successfully completed;

if the writing of said data and the updating of said parity block have both successfully completed, setting said PMD to a value indicating that said parity block and said data blocks are consistent;

if the writing of said data and the updating of said parity block have not both successfully completed, leaving said PMD with said inconsistency-indicating value;

receiving, after said MDU has been updated in the preceding manner, a read request from said processor for one of said data blocks in said MDU;

determining whether the data block to be read is available;

if the data block to be read is available, reading the data block and returning it to said processor;

if the data block to be read is not available, carrying out the following steps;

(a) reading said PMD in said MDU to determine whether said parity block and said data blocks are consistent;

(b) if said PMD in said MDU indicates that said parity block and said data blocks are consistent, carrying out the following steps;

(i) regenerating the unavailable data block from said parity block and data blocks other than the unavailable data block in said MDU; and

(ii) returning the regenerated data block to said processor; and

(c) if said PMD in said MDU indicates that said parity block and said data blocks may be inconsistent, refraining from regenerating the unavailable data block from said parity block and said other data blocks in said MDU.

2. A method according to claim 1, wherein:

(1) said parity block and said set of data blocks are part of a strip distributed across said storage devices;

(2) said MDU comprises said strip and additional similar strips, the consistency of the parity and data blocks in said strips being indicated by said PMD;

(3) one of said storage devices is designated a parity storage device;

(4) said strips are arranged such that the parity block from each strip is on said parity storage device; and

(5) said PMD is also on said parity storage device; and

wherein any of said steps in claim 1 involving reading or writing either said PMD or said parity block comprises the step of accessing said parity storage device.

3. A method according to claim 2, wherein said PMD comprises bits each indicating the consistency between the parity block and the data blocks in a corresponding one of said strips, and wherein each bit, when set, indicates that the parity block and the data blocks in the corresponding strip may be inconsistent and, when cleared, indicates that the parity block and the data blocks in the corresponding strip are consistent.

4. A method according to claim 1, wherein:

(1) said parity block and said set of data blocks are part of a strip distributed across said storage devices;

(2) said MDU comprises said strip and additional similar strips, the consistency of the parity and data blocks in said strips being indicated by said PMD;

(3) said strips are arranged such that parity blocks from different strips generally reside on different ones of said storage devices; and

(4) said PMD is distributed as a PMD segment across all of said storage devices such that the portion of said PMD segment on a given storage device indicates the consistency between the parity block and data blocks for those strips whose parity block resides on that storage device; and

further comprising the step of determining which of said storage devices contains said parity block and said PMD for the strip to which said data is to be written.

5. A method according to claim 4, wherein said PMD comprises bits each indicating the consistency between the parity block and the data blocks in a corresponding one of said strips, and wherein each bit, when set, indicates that the parity block and the data blocks in the corresponding strip may be inconsistent and, when cleared, indicates that the parity block and the data blocks in the corresponding strip are consistent.

6. A method according to claim 1, further comprising the step of sending a write success signal to said processor if the writing of said data has successfully completed.

7. A method according to claim 1, further comprising the step of sending an error signal to said processor if the writing of said data has not successfully completed.

8. A method according to claim 1, further comprising the step of writing the regenerated data block to the location occupied by said unavailable data block in said MDU if said step of regenerating the unavailable data block is carried out.

9. A method according to claim 1, further comprising the step of sending an error signal to said processor if said step of refraining from regenerating the unavailable data block is carried out.

10. A method according to claim 1, wherein said PMD comprises a bit which when set indicates that said parity block and said data blocks may be inconsistent and when cleared indicates that said parity block and said data blocks are consistent.