Monitoring and Changing Engine States

used process instances (and all their associated activities, timers, and process attributes), until it brings the number of memory-resident process instances back within the limit. The engine performs these checks periodically, as specified by the value of the Swap-out Interval.

When a swapped-out process instance is subsequently needed, for example, to change the state of an activity, to handle an expired timer, to evaluate an assignment role, and so forth, the engine finds the process instance in the engine database’s current state tables and swaps it back into memory.

In general, you want the value of the Swap-out Interval to be short enough so that, under heavy load conditions, the number of process instances in memory cannot grow to a point where the available memory resources are exceeded. However, if you make the interval too short, the engine is incurring excessive overhead, swapping more process instances out of and into memory than is needed. Similarly, the value of the Memory-resident Process Limit must be set in accordance with the memory required by executing process instances (which depends on their complexity), the available memory resources, and the value of the Swap-out Interval.

Despite the difficulty of determining the optimum values of each of these configuration options, the combination provides great flexibility in tuning an engine to process the heaviest loads, with the greatest performance, and with the least risk of failure.

NOTE The evaluation of role-based assignment rules—as compared to assignment rules that involve process attributes or linked activities—does not require that the corresponding process instance be resident in memory. Hence, much less swapping is generally involved in executing process instances that employ only role-based assignment rules.

Monitoring and Changing Engine States

After you have started an engine, you should verify that it has started successfully and monitor it periodically. In some situations, you may want to dynamically change the state of engine components.

This section starts by describing how to monitor engines and engine components with the iIS Console, followed by details on how to change engine states. For information on how to perform these operations using Conductor Script, see “Managing iIS Process Engines with Conductor Script” on page 235.

Chapter 4 Managing Engines 117

Page 117
Image 117
Sun Microsystems 3 manual Monitoring and Changing Engine States