Correct! We enforce unique names. We don't use it for any actual logic in Hyperdrive itself, but it makes life way simpler for things like wrangler or the dash
Look at switching to Aurora Serverless v2; it now supports scaling to zero, which is great for side projects that don't need to be on all the time. (Takes <30 seconds to turn back on.)
Should be $43.8/m for 0.5 ACUs (assuming you never scale to zero) + $3.65/m for IPv4. Then there's some extra costs for bandwidth and I/O, but it's pretty small for (most?) workloads.
Yeah, idk what Max is running, but side project, in my mind at least, is something that would fit on a free-tier Supabase/Xata/Neon(also since you probably don't need redundancy or other jazz)
If you bundled many projects together onto a single database, it might be more worth it, but imo it's still a lot more pricy than the hobby alternatives.
Not necessarily? You should be fine for a whole month if you go 0.25 vCPU, 2 GB RAM. Or, too, there is the scale to zero, which again, for a side project, you might not need 24/7 uptime
I sure know every project I have deployed(other than WDOL) has DBs cold-starting, because I definitely don't have enough usage to keep them alive all the time. YMMV may vary of course, depending on your project
Well, we get to cheat a little, with the runtime-generated connection strings.
We probably could if we wanted to. Not saying it's something we'll do, but I'm curious and have never considered that before. What would be the use case for something like that?
Another topic, I think there is a bug in delete hyperdrive feature: after delete a hyperdrive, I can't create another hyperdrive with the same name, because it's already exists.