Optimizing SQL Performance With Parallelism In YugabyteDB YSQL
YugabyteDB YSQL enables parallel scanning of rows with pushdowns for filters and aggregations on hash-sharded tables.
When you connect to YugabyteDB YSQL (the PostgreSQL compatible API) your session executes the SQL statement from one process (the PostgreSQL backend parsing the query and executing the plan) which reads and writes, by batch of rows, from/to the tablet servers where the rows are distributed. In some cases, it makes sense to send multiple batches of operations at the same time, so that they are processed in parallel in multiple tablet servers. In some other cases, when those operations return a lot of rows for example, this may not be efficient because those rows will be finally processed by a...