Are there baseline CPU and Memory requirements for self-hosting standalone Next.js?

[jplwood]

Summary

When self-hosting Next.js, as described in Deploying: Self-Hosting, are there baseline expectations for CPU and Memory for the hosting environment?

My app's hosting env was defaulting to .25vCPU and .5gb memory. The app was incredibly slow (TTFB consistently >2s). I bumped it way up just for testing and suddenly the app's TTFB 75th percentile is at ~500ms. That said, I'm curious if there is an optimal minimum to apply to each container to support Nextjs at baseline?

Additional information

I'm running Next.js in standalone mode in AWS ECS Fargate, but this question applies to any self hosting.

[rootz491]

I'm also wondering about the same. Are there any recommended specs?
Also has anybody done any benchmarks on nextjs self-hosted yet? I couldn't find anything really.

[greenwookez]

What I've observed so far - I have a basic website with NextJS + PayloadCMS under the hood packed into a Docker container. The website itself isn't too large, I mean 10-15 static pages + 5 dynamic ones. And it's still in development phase, so no real load yet, just occasional QAs, so...

It feels very hard on RAM, while barely putting any load on CPU - which is pretty straightforward.
I stay on 1vCPU 3.3Ghz, only because my IaaS provider doesn't allow less. I think I'd go with less CPU if I could.
As of RAM, I'm currently on 2gb, and it shows that 0.6gb is used and 1.7gb is used with cache.

So I think I'd probably need to upgrade to somewhere near 4gb before going into production.