Organizing pools
There are several ways of organizing your pools.
About this task
- If possible, restrict each pool to a single target, but specify
as many nodes as you believe you need to satisfy concurrent access
to the target. The reasons for taking this approach are:
- It avoids the need for the front-end application to specify a target.
- It makes it easier to avoid duplicate connection definitions.
- Because a connection is created for every node-target combination within a pool, having large numbers of both nodes and targets within the same pool may generate more resources than are required.
- The overhead associated with a pool is very small. Therefore there is no reason not to define many pools.
- The expected concurrent usage of each target may be different. If you have more than one target in the pool, it becomes difficult to estimate the number of nodes required.
- You can define a pool containing only one node and one target. This lets a FEPI application allocate a specific session, which is necessary if the target system associates any special qualities with a particular terminal ID. You can use the XSZBRQ global user exit to control access to the pool.
- You can define pools that use different nodes to reference the same target. By making each pool available to a different group of users, you can eliminate competition for resources. Alternatively, you could use each pool to support a different set of properties, according to application requirements.
- If you plan to use the z/OS® Communications Server VTAM CLSDST(PASS) command, other considerations might apply. See Handling CLSDST(PASS).
Do not use names starting with “DFH” for pools.