PostgreSQL 10 is on its way

A multitude of features and fascinating details.

The next major release of PostgreSQL is on its way. It's a huge release in terms of features, improvements and bugfixes.

First things first

Versioning has been changed to a simpler scheme using only two numbers: major and minor version. So, this will be PostgreSQL 10 followed by 10.1 and 10.2 and so on. Okay, now let's get to the real deal 😉

Logical Replication

The inclusion of pglogical as logical replication has been completed. So, PostgreSQL 10 will feature a publish-subscribe replication system which finally supports use-cases like zero-downtime upgrades, data consolidation à la data warehousing and scaling out read load; using one database technology only. I personally find this an important feat for the opensource project!

Parallel Support

Parallel support has been extended far beyond the initial parallel-enabling architecture (e.g. flagging parallel-safe functions) and its first implementation for sequential scans with hash join, nested loops and aggregate functions. Now PostgreSQL includes parallel btree index scans, parallel bitmap heap scans, parallel merge joins and parallel uncorrelated subqueries as well.

Hash Indexes

Finally, hash indexes are going to be fully featured citizens of PostgreSQL, thus the age-old warning against their usage in production is going to be removed.

Fulltext meets JSON

Other improvements will be fulltext searches performed on JSON datasets, plus more utility functions for JSON data. Basically the cornerstone of what almost all modern non-trivial websites need to do to provide a streamlined user experience combined with short development cycles.


In a series of upcoming posts, I am going to cover the most interesting aspects in detail by looking at specific commit sets.

Sven