Method and system for load balancing a distributed database providing object-level management and recovery

• US 9,612,921 B2
• Filed: 02/28/2014
• Issued: 04/04/2017
• Est. Priority Date: 03/01/2013
• Status: Active Grant

• Current Assignee: Teradata Corporation
• Original Assignee: Teradata US, Inc. (Teradata Corporation)
• Inventors: Ferguson, Gregory Paul
• Primary Examiner(s): Leibovich, Yair
• Application Number:
  US14/193,266
• Publication Number:
  US 20140250326A1
• Time in Patent Office: 1,131 Days
• Field of Search: 714 16, 714 20
• CPC Class Codes: G06F 11/1471 involving logging of persistent data for recovery
• CPC Class Codes: G06F 2201/80 Database-specific techniques

Abstract

A method and system for managing operational states of database tables within a multiple-database system. If a particular user session issues a query against a target table that causes a data inconsistency, the target table transitions into an errant state and the session will become interrupted. This errant state is then propagated onto any other table associated with the user session. A session-level recovery process can thereafter be executed to repair and restore database tables associated with the interrupted user sessions without the need to take an entire database system offline.

Claims

1. A computer-implemented method for recovering from database system errors, said method comprising the steps of:

maintaining, within said database system, table states for each database table within said database system;

setting the table state for a database table to an interrupted state upon encountering an error during a database operation affecting said database table, said database operation comprising a query issued during a user session that causes said database table to transition into said interrupted state;

setting the table state for at least one additional database table affected by a query issued during said user session to an interrupted state;

maintaining a recovery log containing a sequential list of database operations issued to said database system;

placing a checkpoint marker in said recovery log indicating a starting point for said restore operations, said a checkpoint marker being placed in said recover log at periodic intervals;

executing a restore operation to restore said database table having an interrupted state to an active state; and

executing a restore operation to restore said at least one additional database table having an interrupted state to an active state, wherein the execution of said restore operations to restore said database table having an interrupted state and said at least one additional database table having an interrupted state to active states includes the steps of;

identifying database operations listed in said recovery log which are associated with said user session; and

re-executing said database operations listed in said recovery log which are associated with said user session and were entered into said recovery log following said checkpoint marker.

2. A computer-implemented method for recovering from database system errors, said method comprising the steps of:

maintaining, within said database system, table states for each database table within said database system;

setting the table state for a database table to an interrupted state upon encountering an error during a database operation affecting said database table, said database operation comprising a query issued during a user session that causes said database table to transition into said interrupted state;

setting the table state for at least one additional database table affected by a query issued during said user session to an interrupted state; and

executing a restore operation to restore said database table having an interrupted state to an active state; and

executing a restore operation to restore said at least one additional database table having an interrupted state to an active state; and

wherein;

the step of executing a restore operation to restore said database table having an interrupted state to an active state, and the step of executing a restore operation to restore said at least one additional database table having an interrupted state to an active state are attempted at periodic intervals.

3. A computer-implemented method for recovering from database system errors, said method comprising the steps of:

maintaining, within said database system, table states for each database table within said database system;

setting the table state for a database table to an interrupted state upon encountering an error during a database operation affecting said database table, said database operation comprising a query issued during a user session that causes said database table to transition into said interrupted state;

setting the table state for at least one additional database table affected by a query issued during said user session to an interrupted state; and

executing a restore operation to restore said database table having an interrupted state to an active state; and

executing a restore operation to restore said at least one additional database table having an interrupted state to an active state; and

wherein;

the step of executing a restore operation to restore said database table having an interrupted state to an active state, and the step of executing a restore operation to restore said at least one additional database table having an interrupted state to an active state occur in response to a system restart.

4. A computer-implemented method for recovering from database system errors, said method comprising the steps of:

maintaining, within said database system, table states for each database table within said database system;

setting the table state for a database table to an interrupted state upon encountering an error during a database operation affecting said database table, said database operation comprising a query issued during a user session that causes said database table to transition into said interrupted state;

setting the table state for at least one additional database table affected by a query issued during said user session to an interrupted state; and

executing a restore operation to restore said database table having an interrupted state to an active state; and

executing a restore operation to restore said at least one additional database table having an interrupted state to an active state; and

wherein;

the step of executing a restore operation to restore said database table having an interrupted state to an active state, and the step of executing a restore operation to restore said at least one additional database table having an interrupted state to an active state occur in response to a user request.