Remote duplicate database facility featuring safe master audit trail (safeMAT) checkpointing

• US 5,794,252 A
• Filed: 12/16/1996
• Issued: 08/11/1998
• Est. Priority Date: 01/24/1995
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Tandem Computers Incorporated (HP Inc.)
• Inventors: Bailey, Bruce W., Mosher, Malcolm Jr.
• Primary Examiner(s): Lintz, Paul R.
• Application Number:
  US08/767,452
• Time in Patent Office: 603 Days
• Field of Search: 707/202, 395/182.05, 395/182.09, 395/182.14
• US Class Current: 707/610
• CPC Class Codes: B65H 1/06 for separation from bottom of pile
• CPC Class Codes: B65H 1/26 with auxiliary supports to facilitate introduction or renewal of the pile
• CPC Class Codes: B65H 1/28 compartmented to receive piles side-by-side
• CPC Class Codes: B65H 2301/42322 from bottom of the pile
• CPC Class Codes: B65H 3/22 by needles or the like engaging the articles
• CPC Class Codes: G06F 11/1451 by selection of backup contents
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 11/1662 the resynchronized component or unit being a persistent storage device re-synchronization of failed mirror storage G06F11/2082; rebuild or reconstruction of parity RAID storage G06F11/1008
• CPC Class Codes: G06F 11/2066 Optimisation of the communication load
• CPC Class Codes: G06F 11/2069 Management of state, configuration or failover
• CPC Class Codes: G06F 11/2074 Asynchronous techniques
• CPC Class Codes: G06F 11/2094 Redundant storage or storage space G06F11/2056 takes precedence
• 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 11/3476 Data logging G06F11/14, G06F11/2205 take precedence
• CPC Class Codes: G06F 2201/80 Database-specific techniques
• CPC Class Codes: G07B 3/04 from a stack
• CPC Class Codes: Y10S 707/915 Image
• CPC Class Codes: Y10S 707/966 Distributed
• CPC Class Codes: Y10S 707/99953 Recoverability

Abstract

A local computer system has local database, application programs that modify the local database, an a transaction manager that stores audit records in a local audit trail reflecting those application program modifications to the local database. A remotely located computer system has a backup database. A remote data duplication facility (RDF) is partially located in the local computer system and partially in the remote computer for maintaining virtual synchronization of the backup database with the local database. The RDF includes an extractor process executed by said local computer system, and a receiver process and a plurality of updater processes executed by the remote computer system. The extractor process extracts audit records from a local audit trail and transmits those records to the receiver process. The receiver process distributes the audit records into one or more image trail files that are associated with a particular updater process. Each updater process reads the audit records in its assigned image trail file and initiates redo operations of database modifications denoted in at least a subset of the audit records against the backup database. Each image trail file has a corresponding image trail buffer that is used to accumulate a number of audit records before they are stored in the corresponding image trail file. Before an updater process applies an audit record against the backup database, it requests from the receiver process the status of the transaction associated with the audit record. Status audit records are stored in a transaction status table as well as in a master image trail file. The RDF system herein described provides a method and system for assuring that each audit record is preserved in the system in the event a failure occurs to any of the processes in the RDF.

Claims

1. A distributed computer database system, comprising:

a local computer system having a local database stored on local memory media, application programs that modify the local database, and a transaction manager that stores audit records in a local audit trail reflecting those application program modifications to the local database;

each audit record having an associated audit trail position in said local audit trail;

a remote computer system, remotely located from the local computer system, said remote computer system having a backup database stored on memory media associated with the remote computer system;

a communication channel for sending messages between said local computer system and said remote computer system; and

a remote data duplication facility, partially located in said local computer system and partially located in said remote computer system, for maintaining virtual synchronization of said backup database with said local database, including;

an extractor process executed by said local computer system that extracts audit records from said local audit trail and transmits said extracted audit records to said remote computer system, a first subset of said audit records denoting updates to said local database, each audit record of said first subset having a transaction identifier indicating an associated transaction, each audit record of a second subset of said audit records indicating a status of one of said transactions;

a receiver process executed by said remote computer system that receives said audit records transmitted by said extractor process and distributes said audit records to one or more image trail buffers in said remote computer system, each image trail buffer associated with an image trail file;

one or more updater processes executed by said remote computer system, wherein each updater process reads said audit records in an assigned one of said image trail files and initiates redo operations of updates specified in said audit records against an assigned portion of said backup database;

said image trail buffers having at least one master image trail buffer and one or more auxiliary image trail buffers, each of said auxiliary image trail buffers including audit records of said first subset, said master image trail buffer including audit records of said second subset;

each of said updater processes including instructions to request a specified status audit record prior to initiating redo operations associated with a corresponding audit record; and

said receiver process including instructions to flush each of said auxiliary image trail buffers to each of said respective image trail files before said master image trail buffer is flushed to its associated image trail file and to store in a memory a safeMATposition denoting an audit trail position of a last status audit record stored in said image trail file associated with said master image trail buffer.

2. The system of claim 1,

wherein said receiver process further includes instructions for storing one or more status audit records in a transaction status table, said transaction status table used to retrieve one or more status records in response to said status requests;

and

wherein said receiver process further includes instructions to respond to a status request associated with an update audit record from one of said updater processes by searching for a status audit record stored in said transaction status table whose transaction identifier matches said requested audit record'"'"'s transaction identifier, said receiver process transmitting said matched status audit record to said requesting updater process once said safeMATposition equals or exceeds said requested audit record'"'"'s audit trail position.

3. The system of claim 2,

wherein said receiver process further includes instructions to transmit in response to said status request additional status audit records whose corresponding audit trail position exceeds said requested update audit record'"'"'s audit trail position and is less than said safeMATposition.

4. The system of claim 1, wherein

each of said image trail files includes one or more blocks of audit records, each block having a block header, each of said image trail files having an end of file position;

said receiver process further includes instructions to determine a timestamp for each of said image trail files when said image trail file is created;

said receiver process further includes instructions to store said timestamp in each block header of said respective image trail file when one or more audit records are written into a corresponding block; and

said receiver process includes instructions to set said end of file position of each of said image trail files to a position at or before a first block header having a timestamp whose value does not match said created timestamp when a predefined condition occurs.

5. The system of claim 1,

wherein said extractor process includes a plurality of message buffers and instructions for buffering groups of said extracted audit records together in said message buffers and transmitting said message buffers to said remote computer system, each transmitted message buffers having an associated sequence number, wherein sequentially transmitted message buffers have associated sequence numbers that follow a predefined sequence;

said extractor process instructions including instructions for continuing, after transmitting a first one of said message buffers to said remote computer system, to buffer groups of said extracted audit records in other ones of said message buffers and for transmitting said other message buffers to said remote computer system.

6. The system of claim 5,

wherein said receiver process further includes instructions to store an expected message sequence number and a restart audit trail position based on said audit trail positions associated with said audit records in said received message buffer, to receive said message buffers transmitted by said extractor process, to respond to each received message buffer whose associated message sequence number does not match said expected message sequence number by transmitting an error message to said extractor process, and to respond to each received message buffer whose associated message sequence number matches said expected message sequence number by sending a reply message to said extractor process to acknowledge receipt of each said message buffer.

7. The system of claim 6, wherein said extractor process further includes instructions:

to respond to said error message by obtaining said restart audit trail position from said receiver process; and

to extract said audit records from said local audit trail starting at restart audit trail position.

8. The system of claim 1,

wherein said remote data duplication facility further comprises a responder process for storing one or more of said status audit records in a transaction status table and for retrieving one or more status records from said transaction status table in response to said status requests.

9. The system of claim 8,

wherein said responder process further includes instructions to respond to one of said status requests by searching for a status audit record stored in said transaction status table whose transaction identifier matches said requested audit record'"'"'s transaction identifier and for transmitting said matching status audit record to said requesting updater process.

10. The system of claim 9,

wherein said responder process includes instructions to reconstruct said transaction status table from those records stored in said image trail file associated with said master image trail buffer.

11. A memory for storing data for access by programs being executed on a remote data processing system having a database system, said remote data processing system in communication with at least one local data processing system including a local database system that is synchronized with said remote database system, said memory comprising:

a receiver process executed by said remote data processing system that receives a plurality of audit records reflecting application program modifications to said local database system, said audit records stored in a local audit trail located in said local data processing system and having an associated audit trail position in said local audit trail;

said receiver process distributing said audit records to one or more image trail buffers, each image trail buffer having an associated image trail file corresponding to an assigned updater process that reads said audit records and initiates redo operations of database modifications denoted in at least a first subset of said read audit records against an assigned portion of said remote database system, each of said audit records in said first subset having a transaction identifier denoting a corresponding transaction;

said image trail buffers including at least one master image trail buffer associated with a master image trail file including a second subset of said audit records, each audit record of said second subset indicating a status associated with one of said transactions; and

said receiver process including instructions to flush each of said image trail buffers to each of said respective image trail files before said master image trail buffer is flushed to said master image trail file and to store in a memory a safeMATposition denoting an audit trail position of a last status audit record stored in said master image trail file.

12. The memory of claim 11,

wherein said receiver process further includes instructions for storing one or more status audit records in a transaction status table, said transaction status table used to retrieve one or more status records in response to said status requests;

and

wherein said receiver process further includes instructions to respond to a status request from one of said updater processes associated with an audit record of said first subset by searching for a status audit record stored in said transaction status table whose transaction identifier matches said requested audit record'"'"'s transaction identifier, said receiver process transmitting said matched status audit record to said requesting updater process once said safeMATposition equals or exceeds said requested audit record'"'"'s audit trail position.

13. The memory of claim 11, wherein

each of said image trail files includes one or more blocks of audit records, each block having a block header, each of said image trail files having an end of file position;

said receiver process further includes instructions to determine a timestamp for each of said image trail files when said image trail file is created;

said receiver process includes instructions to store said timestamp in each block header of said respective image trail file when one or more audit records are written into a corresponding block; and

said receiver process includes instructions to set said end of file position of each of said image trail files to a position before at or a first block header having a timestamp whose value does not match said created timestamp when a predefined condition occurs.

14. The memory of claim 11, wherein

said receiver process includes instructions to receive a plurality of message buffers transmitted from said local data processing system, each transmitted message buffer including a plurality of said audit records, each transmitted message buffer having an associated sequence number, wherein sequentially transmitted message buffers have associated sequence numbers that follow a predefined sequence;

said receiver process includes instructions to respond to each received message buffer whose associated message sequence number does not match an expected message sequence number by transmitting an error message to said transmitting local data processing system;

said receiver process includes instructions to respond to each received message buffer whose associated message sequence number matches said expected sequence number by updating said expected message sequence number in accordance with said predefined sequence; and

said receiver process includes instructions to store a restart audit trail position based on said audit trail positions associated with said audit records in said received message buffer and transmits said restart audit trail position to said local data processing system when so requested.

15. The memory of claim 11,

wherein each of said updater processes requests a specified status audit record prior to initiating said redo operations associated with a corresponding update audit record;

and

said memory including a responder process for storing one or more of said status audit records in a transaction status table and for retrieving one or more status records from said transaction status table in response to said status requests.

16. The memory of claim 15,

wherein said responder process includes instructions to respond to a status request associated with an audit record of said first subset by one of said updater processes by searching for a status audit record stored in said transaction status table whose transaction identifier matches said requested audit record'"'"'s transaction identifier and for transmitting said matching status audit record to said requesting updater process.

17. The memory of claim 16,

wherein said responder process includes instructions to reconstruct said transaction status table from those records stored in said master image trail file.

18. A method for operating a distributed computer system including a local computer system having a database system, said local computer system in communication with a remote computer system having a backup database system at least partially replicating said local database system, said method comprising the steps of:

receiving a plurality of audit records extracted from an audit trail located in said local database system, said audit records reflecting application program modifications to said local database system, each audit record having an associated audit trail position in said audit trail, a first subset of said audit records denoting updates made to said local database system and including a transaction identifier denoting an associated transaction, each audit record of a second subset of said audit records denoting a status of one of said transactions and having a transaction identifier denoting said transaction;

distributing each of said update audit records to one of a set of one or more image trail buffers, each image trail buffer associated with an image trail file assigned to at least one updater process associated with said remote computer system, each said updater process reading audit records stored in said assigned image trail file and initiating redo operations specified in said first subset of audit records against a corresponding portion of said backup database system;

distributing each of said status audit records to a master image buffer associated with a master image trail file in said remote computer system;

prior to flushing said status audit records in said master image buffer to said master image trail file, flushing said update audit records in each of said image trail buffers to said respective image trail file; and

updating a safeMATposition with an audit trail position of a last stored status audit record written to said master image trail file.

19. The method of claim 18,

updating a transaction status table stored in said remote computer system with one or more of said status audit records stored in said master image trail buffer;

receiving a request for a status of a specified update audit record from one of said updater processes;

responding to said status request by searching for one of said status audit records in said transaction status table matching said specified update audit record'"'"'s transaction identifier; and

returning said matched status audit record when said safeMATposition equals or exceeds said specified update audit record'"'"'s audit trail position.

20. The method of claim 19,

returning one or more status audit records from said transaction status table whose audit trail positions exceed said specified update audit record'"'"'s audit trail position and is less than said safeMATposition.

21. The method of claim 19,

restoring contents of said transaction status table when a prescribed condition occurs from status audit records stored in said master image trail file whose audit trail positions are less than said safeMATposition.

22. The method of claim 18, wherein said receiving step further comprises the steps of:

receiving a plurality of message buffers from said local computer system, each message buffer including at least one audit record and having an associated sequence number, wherein sequentially transmitted message buffers have associated sequence numbers that follow a predefined sequence;

storing an expected sequence number based on said associated sequence numbers of said received message buffers and storing a restart audit trail position based on said audit trail positions associated with said audit records in said received message buffers;

sending a reply message to said local computer system for each transmitted message buffer whose associated message sequence number matches said expected sequence number, said reply message enabling reuse of said message buffer; and

transmitting an error message to said local computer system indicating the need to resynchronize said transmissions at said restart audit trail position for each transmitted message buffer whose associated message sequence number does not match said expected sequence number.

23. The method of claim, 18,

wherein each of said image trail files includes one or more blocks of audit records, each block having a block header, each of said image trail files having an end of file position;

determining a timestamp for each of said image trail files, each timestamp identifying when a respective one of said image trail files was created;

wherein said step of distributing said update audit records further comprises the step of dispersing said update audit records to one or more blocks of said respective image trail buffer, each block header containing said created timestamp;

wherein said step of distributing said status audit records further comprises the step of dispersing said status audit records to one or more blocks of said master image trail buffer, each block header containing said respective created timestamp; and

restoring each of said end of file positions for each image trail file when a prescribed condition occurs, said restoration based on said respective created timestamp located in each block header of said image trail file.

24. The method of claim 23, said restoring step including setting said end of file position of each of said image trail files to a position at or before a first block header having a timestamp whose value does not match said created timestamp when a predefined condition occurs.