Adaptive Server uses two basic rules to reduce the number of required worker processes in an adjusted query plan:
If the optimized query plan specifies a partition-based access method for a table, but not enough processes are available to scan each partition, the adjusted plan uses a serial access method.
If the optimized query plan specifies a hash-based access method for a table, but not enough processes are available to cover the optimized degree of parallelism, the adjusted plan reduces the degree of parallelism to a level consistent with the available worker processes.
To illustrate the first case, assume that an optimized query plan recommends scanning a table’s five partitions using a partition-based table scan. If only four worker processes are actually available at the time the query executes, Adaptive Server creates an adjusted query plan that accesses the table in serial, using a single process.
In the second case, if the optimized query plan recommended scanning the table with a hash-based access method and five worker processes, the adjusted query plan would still use a hash-based access method, but with, at the most, four worker processes.