Similar to sharding, working with a large amount of rows within a geospatial table in postgres, will cause a lot of processing time for a single worker. With the release of postgres 9.6, the server is capable of executing queries which can be processed by multiple CPUs for a faster answer. According to the postgres documentation, depending of the table size and the query plan, there might not be a considerable benefit when implementing a parallel query, instead of a serial query.