The degree of parallelism for a query is the number of worker processes chosen by the optimizer to execute the query in parallel. The degree of parallelism depends on both the upper limit to the degree of parallelism for the query and on the level of parallelism suggested by the optimizer.
Computing the degree of parallelism for a query is important for two reasons:
The final degree of parallelism directly affects the performance of a query since it specifies how many worker processes should do the work in parallel.
While computing the degree of parallelism, the optimizer disqualifies parallel access methods that would require more worker processes than the limits set by configuration parameters, the set command, or the parallel clause in a query. This reduces the total number of access methods that the optimizer must consider when costing the query, and, therefore, decreases the overall optimization time. Disqualifying access methods in this manner is especially important for multitable joins, where the optimizer must consider many different combinations of join orders and access methods before selecting a final query plan.