Deadlock detection and resource timeouts in data sharing environments
The deadlock detection intervals and resource timeout value of your data sharing environment should be based on how deadlock detection and resource timeouts work in data sharing environments.
Recommendations: Quick detection of
deadlocks and timeouts is necessary in a data sharing environment
to prevent a large number of waiters on each system. A large numbers
of waiters can cause much longer wait times for timeouts and deadlocks.
The following are two recommendations to help prevent a large number
of waiters from developing on each system:
- If your non-data-sharing DB2® subsystem has a problem with deadlocks, consider reducing the deadlock time to prevent a long lists of waiters from developing. (If you do not have a problem with deadlocks, you most likely will not have to change any parameters for data sharing.)
- If you have stringent timeout limits that must be honored by DB2, consider decreasing the deadlock
time before moving to data sharing, as illustrated in this example:
Assume that you have set the timeout period for your non-data-sharing DB2 subsystem to 55 seconds because you want the wait time for timeout to be at or before 60 seconds. (This assumes that your deadlock time value is five.) In a data sharing environment, reduce the timeout period to 40 seconds. This makes it more likely that your actual wait time for timeouts is at or before 60 seconds.