I/O free recovery set determination

• US 7,702,660 B2
• Filed: 12/12/2006
• Issued: 04/20/2010
• Est. Priority Date: 12/12/2006
• Status: Active Grant

• Current Assignee: Oracle International Corporation (Oracle Corporation)
• Original Assignee: Oracle International Corporation (Oracle Corporation)
• Inventors: Srihari, Vinay, Yurek, Tolga, Chan, Wilson Wai Shun, Pruscino, Angelo, Wang, Tak Fung
• Primary Examiner(s): Wong; Don
• Assistant Examiner(s): Le; Michael
• Application Number:
  US11/638,058
• Publication Number:
  US 20080140733A1
• Time in Patent Office: 1,225 Days
• Field of Search: 707/202
• US Class Current: 1/1
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 11/2023 Failover techniques
• CPC Class Codes: G06F 11/2035 without idle spare hardware
• CPC Class Codes: G06F 11/2097 maintaining the standby controller/processing unit updated initialisation or re-synchronisation thereof G06F11/1658 and subgroups
• CPC Class Codes: G06F 2201/80 Database-specific techniques
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

In a multi-node database system, a list of data blocks to perform recovery operations on in the event of failure of a given node is dynamically maintained on another node. List is available when the given node fails, and allows the database to be made available more quickly.

Claims

1. A method, comprising:

a first node of one or more processors generating redo log records that include redo records, each redo record of said redo records recording a change to a unit of data of a database;

wherein a multi-node computer system comprises said first node and a second node of one or more processors;

before the first node fails, said second node maintaining a list of units of data changed by certain redo records of said redo records; and

after the first node fails, determining which units of data of said database to perform recovery operations without scanning the certain redo records of said redo records, wherein the determining is performed at least in part by reading said list of units of data.

2. The method of claim 1, wherein in response to detecting a failure of said first node, said second node broadcasting said list to one or more other nodes.

3. The method of claim 2, wherein said one or more other nodes forego requesting locks on certain units of data based at least in part on said list.

4. The method of claim 2, wherein said one or more other nodes determine which particular units of data to request on-demand recovery operations based at least in part on said list.

5. The method of claim 1, wherein said second node maintains said list based at least in part on information in the certain redo records.

6. The method of claim 1,

wherein the steps include said first node transmitting copies of said certain redo records to said second node;

and

wherein said second node maintains said list of units based at least in part on information in said copies.

7. A method, comprising:

a first node of one or more processors in a multi-node computer system generating redo log records that include redo records, each redo record of said redo records recording a change to a data block;

wherein the multi-node computer system comprises said first node and a second node of one or more processors;

before detecting a failure of said first node, a second node maintaining certain data that associates certain data blocks with certain redo records that record changes to said certain data blocks; and

after detecting said failure of said first node, identifying the certain data blocks without scanning the certain redo records, wherein the identifying is performed at least in part by reading said certain data.

8. The method of claim 7, wherein the steps further include:

after detecting said failure, receiving a request to perform on-demand recovery operations on said certain data blocks; and

determining which redo records are required for said on-demand recovery operations based at least in part on said certain data.

9. The method of claim 8,

wherein said certain data includes a hash table with a hash function that maps a data block to an entry in the hash table;

and

wherein each entry of said hash table is associated with one or more redo records that record changes to a data block mapped to said each entry.

10. A volatile or non-volatile machine-readable medium storing one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform:

a first node of one or more processors generating redo log records that include redo records, each redo record of said redo records recording a change to a unit of data of a database;

wherein a multi-node computer system comprises said first node and a second node of one or more processors;

before the first node fails, said second node maintaining a list of units of data changed by certain redo records of said redo records; and

after the first node fails, determining which units of data of said database to perform recovery operations without scanning the certain redo records of said redo records, wherein the determining is performed at least in part by reading said list of units of data.

11. The volatile or non-volatile machine-readable medium of claim 10, wherein in response to detecting a failure of said first node, said second node broadcasting said list to one or more other nodes.

12. The volatile or non-volatile machine-readable medium of claim 11, wherein said one or more other nodes forego requesting locks on certain units of data based at least in part on said list.

13. The volatile or non-volatile machine-readable medium of Claim 11, wherein said one or more other nodes determine which particular units of data to request on-demand recovery operations based at least in part on said list.

14. The volatile or non-volatile machine-readable medium of claim 10, wherein said second node maintains said list based at least in part on information in the certain redo records.

15. The volatile or non-volatile machine-readable medium of claim 10, further causing the one or more processors to perform:

said first node transmitting copies of said certain redo records to said second node; and

wherein said second node maintains said list of units based at least in part on information in said copies.

16. A volatile or non-volatile machine-readable medium storing one or more sequences of instructions which, when executed by one or more processors, causes the one or more processors to perform:

a first node of one or more processors in a multi-node computer system generating redo log records that include redo records, each redo record of said redo records recording a change to a data block;

wherein the multi-node computer system comprises said first node and a second node of one or more processors;

before detecting a failure of said first node, a second node maintaining certain data that associates certain data blocks with certain redo records that record changes to said certain data blocks; and

after detecting said failure of said first node, identifying the certain data blocks without scanning the certain redo records, wherein the identifying is performed at least in part by reading said certain data.

17. The volatile or non-volatile machine-readable medium of claim 16, further causing the one or more processors to perform:

after detecting said failure, receiving a request to perform on-demand recovery operations on said certain data blocks; and

determining which redo records are required for said on-demand recovery operations based at least in part on said certain data.

18. The volatile or non-volatile machine-readable medium of claim 17,

wherein said certain data includes a hash table with a hash function that maps a data block to an entry in the hash table;

and

wherein each entry of said hash table is associated with one or more redo records that record changes to a data block mapped to said each entry.