At this point, we are using litespeed and one by one we are slowly fixing the 503 issues ourselves. But since you raised that specific point, I will provide more info from our part.
OLS works like a charm without anny issues on RC. At this point our only problem is with Enhance 503 requests which is indeed serious issue. But with clients who have dedicated plan and OLS this issues are not present.
And for 503 issues. We've compared runcloud and ehnahce configs and found few things that looks promising to solve the issues. But at this point the tests are research is ongoing still.
One thing I can say that there is time limit that keeps processes alive which results in RAM overflow OR NPROC overflow. Then this was lowered alongside few other things, we stopped seeing the 503 for few selected customer where we did the tests.
Mostly 24 NPROC with 2-3 GB RAM fixed it for many users.
As for 100ms difference - never seen that issue, can't comment. PHPMyAdmin worked fine when opened. We are using NodeJS pre-ehnahcen logic (not current version), so can't comment either.
What I can say is the difference between using RunCloud and Enhance. I am not saying Enhance does not have issues. So take it as that.
