EFFICIENT CHECKSUMS FOR SHARED NOTHING CLUSTERED FILESYSTEMS

• US 20130227379A1
• Filed: 02/23/2012
• Published: 08/29/2013
• Est. Priority Date: 02/23/2012
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Gupta, Karan, Tewari, Renu
• Granted Patent: US 9,128,862 B2
• Time in Patent Office: Days
• US Class Current: 714/807
• CPC Class Codes: G06F 11/1004 to protect a block of data words, e.g. CRC or checksum G06F11/1076 takes precedence; security arrangements for protecting computers or computer systems against unauthorized activity G06F21/00
• CPC Class Codes: H03M 13/09 Error detection only, e.g. using cyclic redundancy check [CRC] codes or single parity bit
• CPC Class Codes: H03M 13/2906 using block codes H03M13/2957 takes precedence

Abstract

Embodiments of the invention relate to efficiently employing checksums for shared nothing clustered filesystems. Tools are provided to compute the checksum in response to a read transaction and to utilize the computed checksum to prevent serving corrupted data. Multiple levels of data replication are provided. The checksum computation functions within the multiple levels and addresses a specified data block that is the subject of the read transaction.

Claims

1. A method comprising:

amplifying a write transaction, the amplification including at least three levels of replication;

issuing a first read transaction to a first data block at a first replication level, including computing a first checksum on a first read block of the first read transaction;

issuing a second read checksum on the first read block of the first read transaction at a second replication level; and

comparing the first checksum and the second checksum, including serving data associated with the first read transaction if a match of the first and second checksums is returned and issuing a third read checksum on the first read block of the first read transaction at a third replication level if the match is not returned.

2. The method of claim 1, wherein the third checksum issuance is limited to a first mismatch of the first and second checksums.

3. The method of claim 2, further comprising comparing the third read checksum to the first and second checksums and serving data associated with one of the first and second read transactions that match the third read checksum.

4. The method of claim 2, further comprising a failed transaction in response to a second mismatch of the third checksum with both the first and second checksums.

5. The method of claim 1, wherein issuance of the read checksum transaction on one of the replication levels is limited to a read path associated with the read transaction

6. The method of claim 1, further comprising storing each of the first, second, and third read checksum data in volatile memory, and in response to a second read transaction for a sub-block of the first data block, reading an associated checksum from the volatile memory for comparison of checksum data.

7. A system comprising:

a processor in communication with memory;

a functional unit in communication with the memory, the functional unit comprising tools to support completion of a read transaction, the tools comprising;

an amplification manager to amplify a write transaction, an associated amplification of the write transaction including at least three levels of replication;

a first read manager to issue a first read transaction to a first data block at a first replication level, including the first read manager to compute a first checksum on a first read block of the first read transaction;

a second read manager to compute a second read checksum on the first read block of the first read transaction at a second replication level; and

a comparison manager in communication with both the first read manager and the second read manager, the comparison manager to compare the first checksum and the second checksum, including the comparison manager to serve data associated with the first read transaction if a match of the first and second checksums is returned and the comparison manager to instruct a third read manager to compute a third read checksum on the first read block of the first read transaction at a third replication level if the match is not returned.

8. The system of claim 7, wherein the third checksum computation by the third read manager is limited to a first mismatch of the first and second checksums.

9. The system of claim 8, further comprising the comparison manager to compare the third read checksum to the first and second checksums and to serve data associated with one of the first and second read transactions that match the third read checksum.

10. The system of claim 8, further comprising a failed transaction in response to a second mismatch of the third checksum with both the first and second checksums.

11. The system of claim 7, wherein issuance of the read checksum transaction on one of the replication levels is limited to a read path associated with the read transaction

12. The system of claim 7, further comprising a storage manager in communication with each of the read managers, the storage manager to store each of the first, second, and third read checksum data in volatile memory, and in response to a second read transaction for a sub-block of the first data block, the storage manager to read an associated checksum from the volatile memory for comparison of checksum data.

13. A computer program product comprising a computer readable non-transitory storage medium having computer readable program code embodied therewith, the computer readable program code when executed on a computer causes the computer to:

amplify a write transaction, the amplification including at least three levels of replication;

issue a first read transaction to a first data block at a first replication level, including computing a first checksum on a first read block of the first read transaction;

issue a second read checksum on the first read block of the first read transaction at a second replication level; and

compare the first checksum and the second checksum, including instructions to serve data associated with the first read transaction if a match of the first and second checksums is returned and instructions to issue a third read checksum on the first read block of the first read transaction at a third replication level if the match is not returned.

14. The computer program product of claim 13, wherein the third checksum issuance is limited to a first mismatch of the first and second checksums.

15. The computer program product of claim 14, further comprising program code to compare the third read checksum to the first and second checksums and instructions to serve data associated with one of the first and second read transactions that match the third read checksum.

16. The computer program product of claim 14, further comprising a failed transaction in response to a second mismatch of the third checksum with both the first and second checksums.

17. The computer program product of claim 13, wherein issuance of the read checksum transaction on one of the replication levels is limited to a read path associated with the read transaction.

18. The computer program product of claim 13, further comprising program code to store each of the first, second, and third read checksum data in volatile memory, and in response to a second read transaction for a sub-block of the first data block, instructions to read an associated checksum from the volatile memory for comparison of checksum data.