Threads are an important DCE feature that can affect DSS performance. It is recommended to conserve threads whenever possible. Although threads are convenient for structuring an application, they have a cost. Each thread has its own stack and associated data structures that add to the memory requirements of the application. In addition, threads are usually scheduled in round-robin fashion, with each thread waiting its turn to run. Thus, the more active threads, the longer a given thread may have to wait before it runs.

Also, creating and destroying threads takes time. Instead of creating a thread whenever needed and destroying it when the job is done, consider using a thread pooling strategy. This way, threads that have finished their assignment check a queue of jobs for more work, and wait if no work is available.

Because a thread can be interrupted at any moment and another one scheduled, try to avoid holding locks unnecessarily. If a thread is interrupted while it holds an important lock and another thread is dispatched, only to try to grab the lock and get blocked, time has been wasted on context switching back and forth. The second thread is not going to make any progress until the first one releases the lock. You can reduce this occurrence by holding locks for shorter times, and by using finer grained locks. Use different locks for logically separate structures rather than one global lock. However, you have to trade this against the cost of locking and unlocking.