What is High Availability to Postgres? I’ve staked my career on the answer to that question since I first presented an HA stack to Postgres Open in 2012, and I still don’t feel like there’s an acceptable answer. No matter how the HA techniques have advanced since then, there’s always been a nagging suspicion in my mind that something is missing.
But I’m here to say that a bit of research has uncovered an approach that many different Postgres cloud vendors appear to be converging upon.
Postgres is one of those database engines that carves out a niche and garners adherents with various levels of religious zeal. The community, while relatively small when compared to that of something like MongoDB, is helpful almost to a fault. Members from the freshest minted newb to the most battle tested veteran will often trip over themselves to answer questions found in the various dedicated forums, mailing lists, and chat rooms.
The Postgres system catalog is a voluminous tome of intriguing metadata both obvious and stupendously esoteric. When inheriting a Postgres database infrastructure from another DBA, sometimes it falls upon us to dig into the writhing confines to derive a working knowledge of its lurking denizens. The trick is to do this before they burst forth and douse us with the database’s sticky innards and it experiences a horrible untimely demise.
Recently on the pgsql-performance mailing list, a question popped up regarding Postgres RAM usage. In this instance Pietro wondered why Postgres wasn’t using more RAM, and why his process was taking so long. There were a few insightful replies, and they’re each interesting for reasons that aren’t immediately obvious. Let’s see what is really going on here, and perhaps answer a question while we’re at it.
Pietro presents several postgresql.conf settings, but here are the ones that matter:
MySQL has had a REPLACE INTO syntax to perform “UPSERT” logic since practically the very beginning. For the longest time, users who wanted to switch to Postgres, but for whatever reason relied on this functionality, were essentially trapped. Postgres 9.5 changed all that, but why did it take so long? As with much of Postgres history, it’s a long story.
To really understand where Postgres started, we need to look at the “old” way of handling a row merge.
