Comment by Escapado
16 days ago
I appreciate the article from a technical perspective. Is there anyone here who has used it in production regardless of the criticism and has some perspective to share on where it actually falls flat or situations where performance degrades massively? Them having built an MMO around it makes it look like the technical limitations around this pattern might(!) only matter for very specific workloads or for poorly implemented reducers - and my gut feeling is that you can also reach for postgres or sqlite and write very inefficient queries that quickly bog your system down as well. Just curious how it stacked up in the real world.
I am the SpacetimeDB developer. Several assertions in this article are simply wrong. You can read our response here: https://spacetimedb.com/blog/benchmarking
I’ve been following for a bit, but have intentionally avoided using it. It’s prob fine for vibe coders (increasingly seems to be their target) and proof of concept work.
There seems to have been a fair bit feature churn across their whole stack. I’ve seen a fair few projects and tooling fizzle out cuz of the moving targets. It’s not really clear what is long term stable.
There is a lot of over promising and under delivering (features advertised in their 1.0 still not here). They own your whole stack, so deviating from what they provide is an unhappy path. If you need something beyond what is in their current feature set (not what is promised!) then you will probably have an easier time constructing such a system out of the hundreds of open-source components built for modular construction.
Their hosting has not been stable (no LTS either, so you are forced to live on the edge of their deploy schedule). Limited hosting options (only US iirc). Migrations and long term management don't seem to be in a great state either (external tooling needed). Trying to escape their stewardship and selfhost leaves you with a restrictive license…
It seems they are trying to follow in Michael Stonebrakers footsteps, first with copying VoltDB and now DBOS. It’s not clear what their priorities are, when they haven’t yet delivered a single solid product yet.
I do not really see how their business model is any different than VoltDB which was also open source with gated features. The VoltDB community died off for this reason, not sure why we would expect anything different to happen here. Unless they change their license, I would not expect a thriving community like Postgres or SQLite/LibSQL