Choose Postgres queue technology

> Over-engineering is as incorrect as under-engineering.

I know what you mean. The catch is figuring out 'the correctly sized' answer. Uncertainty is a huge factor here.

I think a lot of projects that are called 'over-engineered' have the benefit of hindsight. At least some of such projects were rationally designed based on what people knew at the time. There is often significant uncertainty. When engineering resources are hard to acquire, the powers-that-be often want to 'get it right' the first time. Translated: over-engineer it, because we don't know when we'll get resources for the next iteration.

I suppose my point is this: the next time somebody finds some 'over-engineered' code, it can be insightful to learn about the project history. (Example: some projects have overly optimistic prediction of usage. This perhaps suggests that pessimism is one antidote to over-engineering.)

That's kind of missing the parent's point. If you wanted to ensure emails arrive, that sounds like another queue that could be backed by a different table that is also produced into as part of the original transaction.

Exactly. A partial index should make things fly here.

During my tenure as CTO at a fintech company I built a banking engine using postgres backed queue system using Elixir / Phoenix. It's still in use today. The company processed large volumes of transactions and we were able to do things in real-time in terms of payments. Our system reached a point where I realized that we can scale almost infinitely just using a 2 tier architecture (Elixir / Phoenix / Oban and PostgreSQL)

The industry standard for real-time was anything under 30s end-to-end. We hit many scaling limits as our platform grew. We worked on improving the system instead of looking at postgresql as a problem. We measured and measured and postgresql was never the problem when it came to scaling. Most of the problems that existed was mostly due to design decisions in the system or some technical debt that needed to be solved. When we solved those problems the system flew.

There was one incident where our PostgreSQL instance CPU usage went up to 100% during peak hours and would cause problems in the system. We wondered why, there was a lot of debate around the topic, PostgreSQL doesn't scale, we need a larger instance, or move to a dedicated queue system etc...

I didn't believe the hypothesis that PostgreSQL didn't scale. So I conducted a performance audit on the system myself I discovered one of the engineers forgot to add an index on one of the columns on a major table with > 60m records. There was a lot of back and forth between engineers that discussed whether we should add the index. I simply conducted a small experiment and PG analyze clearly showed a missing index in one of the key tables. After we added a single line of code and deployed the index, during peak hours DB CPU usage would not even exceed 20% we had to scale down our DB to save some $$.

PostgreSQL backed queue system does work.

I've been building distributed systems in large and small scales for over a decade. Before building systems with Elixir / Phoenix I also used to work with Rails where the default is as mentioned in this article is you just adopt Redis. While Redis does have additional overhead when it comes to management of the infrastructure, it also worked well.

Having been through both I'd say a simple PostgreSQL backed solution is the clear winner when it came to not having to manage a 3rd dependency when you already have a complex system. There was no 3rd or 4th piece of the puzzle to theorize hypothesize or experiment with. You just had 2 pieces it is either your application or your database that's the problem. That simplicity saved us a lot of time for debate and doing experiments also squashed any dogmatic values and opinions that just didn't hold any water.

Nah, afraid not. I wish.

We always wanted to open source it, but we got bought out by a big and very IP protective company before we got the chance.