Partitioning Improvements in PostgreSQL 11
Robert has been involved in the PostgreSQL project since 2008, first as a patch reviewer and committer and a later as a major developer. Prior to that, he was a PostgreSQL application developer for nearly ten years. Features he has worked on include parallelism (9.4-9.6), the replacement of System V shared memory with anonymous shared memory (for 9.3), read and write scalability improvements (for 9.2), index-only scans (with Heikki Linnakangas and Ibrar Ahmed, for 9.2), unlogged tables (for 9.1), and left join removal (for 9.0). He works at EnterpriseDB as Chief Database Architect for the Database Server.
No video of the event yet, sorry!
PostgreSQL 10 was the first release to feature built-in declarative partitioning. While this feature offered significant performance and ease-of-use benefits as compared with previous releases, it left out numerous user-visible features and performance optimizations. PostgreSQL 11 is expected to offer significant further improvements to the table partitioning feature. These include both SQL-visible features, such as hash partitioning, cascaded index creation, and the ability (with limitations) to define foreign keys referencing a partitoned table, as well as performance optimizations such as faster partition pruning, run-time partition pruning, and partition-wise join. In this talk, I'll give a preview of what users can expect from the partitioning feature in PostgreSQL 11 and what remains to be done in PostgreSQL 12 and beyond.
- 50 min
- PostgresConf US 2018
- Postgres Internals