PostgreSQL : Decoding Partition
Beena has about 7 years experience in PostgreSQL and was involved in developing features like multiple synchronous replication, making wal segment size configurable at initdb time, addition of default partition and runtime partition pruning.
No video of the event yet, sorry!
Trigger based methods for partitioning were replaced with native declarative partitioning in PostgreSQL 10 which enhanced the performance. V10 introduced the declarative partition, adding range and list partitions with support for sub partitioning. It has modified the internal tuple routing for partitioning. V11 added hash partitions, default partitions: a catch-all partition for tuples that do not match the declared partitions, improved update by allowing us to move updated tuples to a different partition. Not only did it improve the planner optimisation but also extended the optimisation to the executor.
This talk would cover the different features of partitions in v10 and V11 showing the behaviour and performance difference with the previous versions; finishing off with the new improvements in 12.
- 45 min
- PGConf India, 2019
- Database Engine Developer