High availability for databases
A database must be made highly available to ensure that data is available if a failure occurs. You can introduce a database with features that suit the needs of your organization, based on the level of protection and the associated costs.
A database server can fail as a result of any of the following factors:
- Environmental, such as temperature
- Hardware failure
- Software failure
- Network connectivity interruption
- Human intervention
Different database configurations provide different failover and system protection capabilities. The following table summarizes the attributes of various high availability features for databases. Failover time is based on the optimum cluster configuration and varies with the products that are used:
Database | Availability type | Failover time | Protection level | Additional cost |
---|---|---|---|---|
Warm standby mode with IBM® DB2® and shared disk | Warm standby | Approximately 1 minute | No local protection from disk failure, only from software or server failure. | Db2 license |
Hot standby mode with IBM Db2 High Availability Disaster Recovery | Hot standby | Approximately 30 seconds | Protection from software failure, storage failure, server failure, and site failure. | Db2 license |
Database availability with Db2 automatic client reroute feature | Enhances availability when Db2 or Db2 HADR is installed. | Dependent on whether Db2 or Db2 HADR is installed. | Supplementary support for to Db2 HADR and IBM Db2 with shared disk. Assists with database reconnection and helps to alleviate symptoms of failover. | None |
Fault tolerance and scalability with IBM Db2 pureScale® Feature | Active-active cluster | Instant | Protection from software failure, storage failure, server failure, and site failure. Loads are balanced and failover is instant, due to the active cluster. | IBM Db2 Enterprise Server Edition entitlement |
Database clustering with Oracle Real Application Clusters | Active-active cluster | Instant | Protection from software failure, server failure, and site failure. Loads are balanced and failover is instant, due to the active cluster. | For more information, see the Oracle product information. |