PostgreSQL 10 High Performance - Third Edition

Hash generation

The standard practice for most PL/Proxy implementations is to partition based on the hashtext of the field you're splitting on, which allows splitting among a number of nodes fairly without knowing the distribution of the dataset in advance. hashtext is a PostgreSQL-provided internal function that takes in any text input and generates an integer as output with a hash code. If you AND the result at the bit level, to only take the lowest few bits, this turns out to be a very quick and fair way to distribute load among multiple systems. For this to work, the number of partitions needs to be a power of 2 (2, 4, 8, 16, and so on) and then you bitwise AND against one less than that number. So for two partitions that's & 1, for four ...