3 ms·

I used it a year ago and had to move off, just too many errors, a few seemingly lost deployments and needing in general to reconnect or turn things off/on to ge

I used it a year ago and had to move off, just too many errors, a few seemingly lost deployments and needing in general to reconnect or turn things off/on to get them to work. Definitely felt very beta. Final straw though really was testing DB. I had a $40/mo dedicated server and I spun up their recommended few-node cluster for postgres. Query response time was something like 5x faster for the dedicated server vs their similarly priced setup. I tried upgrading the the top of the line, still much slower and at that point many multiples more expensive. It wasn't just that though, the entire app was sluggish, whereas locally or with a dedicated box it felt incredibly snappy. I'd have had to be spending something like ~2k/mo to get their top of the line nodes across every service and still would have to accept half the speed of my entire app. The edge isn't very useful if it's not powerful! Disclosure: I work at Vercel, and I do like what Fly.io does generally. Had these opinions well before working at Vercel was even a consideration. I think a lot of serverless/edge type hosts are hiding their true cost behind cheap low powered nodes. Especially if the most powerful nodes are still less powerful than a very mid-tier dedicated box, there goes your entire app performance.

napsterbr 4y ago
To be fair, this sounds like an apples to oranges comparison. Given any "cloud" instace (VM), it's pretty much guaranteed it will be considerably slower than a dedicated server of same size. Also, it seems you are comparing a local database (single dedicated server) with a clustered database ("few-node cluster"). That said, I totally get the cost-benefit argument and this is also why I started using dedicated servers for my own projects.
- nwienert 4y ago
  The point is it's actually impossible to get good performance, even when I scale the nodes up to their highest the raw CPU is a lot lower, and that really matters. Also the cluster has nothing to do with it as this was testing reads and clusters should be basically the same as all they add is a load balance to any one node. Not to mention I tried every combo: just one server vs two or three. Nothing made a difference. The diff was huge at the time, something like 3x.
Laaas 4y ago
For many (if not most) use cases, the latency is more important than the compute.
- nwienert 4y ago
  Definitely not true for my very common use case of Postgres + Hasura + Node API. Especially the DB is way faster based on compute. You're talking ~300+ms gains in perf just in the DB layer, with the graph and node layers also getting significant bumps. I was seeing a single dedicated host in Virginia be way, way faster (like 80% faster reply times) than the top of the line cluster that was in Phoenix (at the time very close to me). If you are serving a page that needs to do a few joins across a non-trivial dataset that's all it requires to immediately tilt the equation completely back to raw CPU being the dramatic deciding factor in speed.

napsterbr 4y ago

To be fair, this sounds like an apples to oranges comparison. Given any "cloud" instace (VM), it's pretty much guaranteed it will be considerably slower than a dedicated server of same size. Also, it seems you are comparing a local database (single dedicated server) with a clustered database ("few-node cluster"). That said, I totally get the cost-benefit argument and this is also why I started using dedicated servers for my own projects.

nwienert 4y ago
The point is it's actually impossible to get good performance, even when I scale the nodes up to their highest the raw CPU is a lot lower, and that really matters. Also the cluster has nothing to do with it as this was testing reads and clusters should be basically the same as all they add is a load balance to any one node. Not to mention I tried every combo: just one server vs two or three. Nothing made a difference. The diff was huge at the time, something like 3x.

Laaas 4y ago

For many (if not most) use cases, the latency is more important than the compute.

nwienert 4y ago
Definitely not true for my very common use case of Postgres + Hasura + Node API. Especially the DB is way faster based on compute. You're talking ~300+ms gains in perf just in the DB layer, with the graph and node layers also getting significant bumps. I was seeing a single dedicated host in Virginia be way, way faster (like 80% faster reply times) than the top of the line cluster that was in Phoenix (at the time very close to me). If you are serving a page that needs to do a few joins across a non-trivial dataset that's all it requires to immediately tilt the equation completely back to raw CPU being the dramatic deciding factor in speed.