Thanks for this commentary! At a startup where we are preparing to shard Postgres. I'd be curious if you're familiar with AWS limitless, and how you would have approached deciding whether to use it vs. the approach in the blog post had it existed back in 2021.
I’m a solid “no” on any primary datastore database thingy that’s under 5 years of industry wide production workload experience, and would seriously weigh it against something with 10+ years industry use.
In 2019 when I was interviewing at companies for my next position I heard from a few places that the original Aurora for Postgres lost their data. It seems like the sentiment on mainline Aurora has improved a bit, but I would never bet my company’s future on an AWS preview technology. Better the devil you know (and everyone else knows).
My biggest concern with Limitless – other than inherent performance issues with Aurora – is that according to their docs, it’s built on Aurora Serverless. IME, Serverless anything tends to get absurdly expensive very quickly.