The time slice configuration parameter prevents tasks that execute on the RAPCache database from monopolizing engines.
Each task is allowed to execute on an engine for a maximum amount of time equal to the sum of the time slice and cpu grace time values. The default value for the time slice parameter is 100 clock ticks, and 500 clock ticks for the cpu grace time parameter.
Each time a task comes to a yield point, a check is performed to see if time slice has been exceeded. If execution time does not exceed time slice, the task continues to execute. If the time slice has been exceeded, the task voluntarily relinquishes the engine within the cpu grace time interval, and the next task that is waiting for the engine starts to execute.
A task may encounter a yield point and exceed the time slice interval. This occurs, for example, if the task is performing I/O activities and the system is using a slow hard drive. When the cpu grace time ends, the RAPCache database terminates the task with a time slice error. If you encounter time slice errors, increase the value of cpu grace time. See the Adaptive Server Enterprise Performance and Tuning Series: Basics.