When planning your data sharing configuration for the high availability, the primary concerns are the physical protection of the coupling facility, the structures in the coupling facility, and network connectivity. The SCA and lock structure are both necessary for the data sharing group to function.
About this task
To plan for high availability of your data sharing group, take the following actions:
Procedure
- Duplexing the group buffer pool and store the SCA and lock structures in failure-isolated coupling facilities as described in Planning for coupling facility availability for data sharing.
- Define a dynamic virtual IP address (DVIPA) for the Db2 group and a DVIPA for each Db2 member in the group.
DVIPA addressing gives you the ability to assign a specific virtual IP address to a data sharing group and to each member of the group. This address is independent of any specific TCP/IP stack within the Parallel Sysplex®. Even if a member is moved to another z/OS® system, as in the case of a failure or maintenance, the member remains accessible and retains the same virtual IP address.
- To achieve the highest level of application availability at the database level, route connections using the Db2 Universal JDBC Driver or Db2 Connect Server with connection concentrator and Sysplex Workload Balancing support enabled. To route all connections through a Db2 Connect Server with Sysplex support enabled, you must first configure the Db2 group and each Db2 member to use DVIPA. The Db2 Connect Server then manages the connections to the group using the DVIPAs.
- To support JTA and JTS distributed transactions, the Db2 Universal JDBC Driver and the WebSphere® Application Server can be used to coordinate commits against a data sharing group. If using the Db2 Universal JDBC Driver, group and member DVIPA need to be configured in order to recover indoubt threads after a system failure.