Partitioning tables in Postgres can be an extremely risky endeavor. Unfortunately on many larger systems, it’s also essentially a requirement; the maximum size of a Postgres table is 32TB. This isn’t just because converting an existing table to a series of partitions is expensive or time consuming. We must consider how the query planner will react to the partitioned version of a table. There’s also the very real risk we will (or already have) implement flaws in the trigger or constraint logic.
One of the things Postgres has been “missing” for a while is logical replication based on activity replay. Until fairly recently, in order to replicate single tables from one database to another, we had to encumber the table with performance-robbing triggers coupled to a third party daemon to manage transport. Those days might finally be behind us thanks to pglogical.
But is it easy to use? Let’s give it a try on our trusty sensor_log table.
For all of those warehouse queries that never seem to complete before the heat death of the universe, there’s often a faster version. Sometimes this is due to a fundamental misunderstanding of how queries work, or how Postgres specifically functions. The trick is knowing when to back away slowly from an ugly but efficient query, and when to inject a flurry of predicates to fully illustrate the original intent of the query so the planner makes better decisions.
Timing can often be extremely fortuitous. Yesterday marked the official release of Postgres 9.6!
Yaaaaayy…
I’ve covered 9.6 previously, but that was a beta and clearly doesn’t count. Besides, while the beta was undoubtedly high quality, the frequency of patch turnover is enough to produce a significantly different final release. So let’s skim through the release notes a bit for stuff that really stands out or seems different from last time.
When it comes to putting Postgres through its paces, we often turn to benchmarks to absolutely bury it under a torrent of oppressive activity. It’s a great way to obtain maximum performance metrics and also observe how Postgres reacts and breaks down under such pressure. But these kinds of tests aren’t really practical, are they? After all, many such simulated workloads are nothing but bragging rights measured against previous Postgres releases, or for hardware comparisons.
