Automated method for identifying and repairing logical data discrepancies between database replicas in a database cluster

• US 8,126,848 B2
• Filed: 09/14/2009
• Issued: 02/28/2012
• Est. Priority Date: 12/07/2006
• Status: Active Grant

• Current Assignee: Robert Edward Wagner
• Original Assignee: Robert Edward Wagner
• Inventors: Wagner, Robert Edward
• Primary Examiner(s): ALAM, SHAHID AL
• Application Number:
  US12/558,922
• Publication Number:
  US 20100005124A1
• Time in Patent Office: 897 Days
• Field of Search: 707/609, 707/610, 707/640, 707/682, 707/648, 707/691, 707/705, 707/803, 709/217, 709/223, 719/318
• US Class Current: 707/648
• CPC Class Codes: G06F 16/27 Replication, distribution or synchronisation of data between databases or within a distributed database system; Distributed database system architectures therefor

Abstract

A method and system for monitoring and maintaining the consistency of replicated databases in a shared-nothing database cluster architecture is presented. The method involves the continuous monitoring of the database transaction logs that are maintained by the autonomous database managers that manage the individual database replicas in the cluster. In the event that data discrepancies are detected via the comparison of the transaction logs of the individual database replicas, remedial action is carried out according to configured rules in the main cluster controller. These remedial actions may include, but are not limited to, making a database node inactive within the cluster, or automatically repairing any data discrepancies that are detected among the database replicas in the cluster.

Claims

1. A system for monitoring and maintaining data consistency comprising:

a plurality of database nodes each capable of operating independently each containing at least one database;

a transaction log for recording each insertion deletion or modification of data for each of the at least one databases in the database nodes;

a controller node in communication with the database nodes that can receive a series of database transaction requests from a client and that sends the series of database transaction requests to the plurality of database nodes and monitors the results of each of the series of database transaction request at each of plurality of database nodes for inconsistencies and when inconsistencies are detected for a database transaction request, determining the change made by the database transaction request in each of the plurality of database nodes by i) identifying the change made by the majority of the database nodes;

ii) making each of the plurality of database nodes identified as not consistent with the majority, inactive;

ii) applying changes to in each of the plurality of database nodes not conforming to the majority to cause the change made in the majority of the database nodes to be made in each of the plurality of database nodes not conforming to the majority;

iv) making each of the plurality of the database nodes previously identified as not consistent with the majority, active;

whereby the database nodes are kept consistent.

2. The system of claim 1 wherein the controller node identifies the change made by one of:

a) the active database node that first completed the transaction request,

b) the active database node that has completed the most transaction requests, and

c) the active database node that is the most authoritative,

when the change made by the majority of the database nodes can not be determined.

3. A method for monitoring and maintaining data consistency between a plurality of database nodes each containing at least one database, comprising the steps of:

a) receiving at least one database transaction request from at least one client node;

b) applying the at least one database transaction request from the at least one client node to all of the database nodes;

c) monitoring the plurality of database nodes and detecting inconsistencies in the results of each of the at least one database transaction request between the plurality of database nodes;

d) when inconsistencies are detected for a database transaction request, determining the chance made by the database transaction request in each of the plurality of database nodes by

i) identifying the change made by the majority of the database nodes;

ii) making each of the plurality of database nodes identified as not consistent with the majority, inactive;

ii) applying changes to in each of the plurality of database nodes not conforming to the majority to cause the change made in the majority of the database nodes to be made in each of the plurality of database nodes not conforming to the majority;

iv) making each of the plurality of the database nodes previously identified as not consistent with the majority, active.

whereby the database nodes are kept consistent.

4. The method of claim 3 further comprising completing step iv) if each of the plurality of the databases nodes previously identified as not consistent with the majority has applied transaction requests applied by all the other of the plurality of the database nodes.

5. The method of claim 3 further comprising recording the inconsistency detected and the changes applied to each of the database nodes identified as not conforming to the majority.

6. The method of claim 3 wherein the step of identifying the change made by the majority of the database nodes is replaced by identifying the change made by one of:

a) the active database node that first completed the transaction request,

b) the active database node that has completed the most transaction requests, and

c) the active database node that is the most authoritative.

7. The method of claim 3 wherein the step of identifying the change made by the majority of the database nodes is replaced by the step of identifying the change made by one of:

a) the active database node that first completed the transaction request,

b) the active database node that has completed the most transaction requests, and

c) the active database node that is the most authoritative,

when the change made by the majority of the database nodes can not be determined.

8. The method of claim 3 further comprising the step of, when inconsistencies are detected that do not correspond to any database transaction request,

i) identifying the value of the inconsistent data in the majority of the database nodes;

ii) making each of the plurality of database nodes identified as not consistent with the majority, inactive;

iii) applying changes to in each of the plurality of database nodes not conforming to the majority to cause the change made in the majority of the database nodes to be made in each of the plurality of database nodes not conforming to the majority; and

iv) making each of the plurality of the database nodes previously identified as not consistent with the majority, active.

9. The method of claim 8 further comprising the step of waiting a period of time after detecting the consistency and before identifying the value of the inconsistent data in the majority of the database nodes.

10. The method of claim 3 where the step of identifying the change made by the majority of the database nodes is replaced by the step of identifying the change made to a master database node.