May 2018
Intermediate to advanced
576 pages
30h 25m
English
By default a query will use only one process. Parallel query is enabled by setting max_parallel_workers_per_gather to a value higher than zero. This parameter specifies the maximum number of additional processes that are available, if needed. So a setting of 1 will mean you have the leader process plus one additional worker process, so two processes in total.
The query optimizer will decide if parallel query is a useful plan based upon cost, just like other aspects of the optimizer. Importantly, it will decide how many parallel workers to use in its plan, up to the maximum you specify.
Across the whole server the maximum number of worker processes available is specified by the max_worker_processes parameter, set at server start ...