Compute times optimizations

So I own a serverless map website and it does a lot of connecting to the database and dropping the connections basically for each time a user loads a part of the map. Due to this my compute time goes up a crazy amount it seems. more than it does on any other service i use. Is there anyway to optimize this? a lot of the Suspend compute Operations take a few seconds and i'm not sure why.

11 Replies

other-emerald•3mo ago

Hey, a good starting point is always to review all your currently active branches and their compute settings. Make sure the min is not too high to avoid over-provisioning. You can also review the CPU graph in the branch monitoring tab. On the free tier, if your min is 0.25 CPU and 0.25 CPU is never exceeded, you can have your database running 24/7 without hitting the limits. I would be surprised if supsend operations would play a role here.

equal-aquaOP•3mo ago

Is there anywhere to see clearly a compute time graph to see what is taking up a decent amount of compute time in my project? my database doesnt even go over .05 of used CPU (it peaked at .07 within the last day). So I honestly dont get why changing the min compute (even though its on its lowest setting) really effects my compute times.

equal-aquaOP•3mo ago

it really seems to me that suspended operations are what is taking up all of my compute time. this really adds up at the end of the day and month to where my compute time runs out of the month and my database goes down.

other-emerald•3mo ago

Check the CPU graph in the monitoring tab. There are two lines: allocated and used CPU. If your database is running, it always has at least the min CPU allocated. It starts with 0.25, scales to 0.5 and so on. Even if you only use 0.05 compute, if your min is at 2 CPU, we'll allocate 2 CPUs and you will use much more compute than needed.

equal-aquaOP•3mo ago