Might as well refer to TechEmpower benchmarks.
https://github.com/just-js/just
It seems to be a much tinier JavaScript runtime than Node.js (still using v8), but linux only
The benchmark is probably unrealistically optimized code but even so, it implies Node.js itself has a large performance overhead
If you pay attention you'll notice that just.js is using postgres as their DB, while all of the node benchmarks are handicapped by either using mongoose/mongodb or MySQL.
There is no node benchmark with postgres, but all of the fastest benchmarks used it.
"lithium" is a good example to show how much of an impact switching to postgres has. All of the 4 lithium benchmarks are identical except in what DB they use.
The results are: lithium-postgres-batch (#2 - 659850), lithium-postgres-beta (#13 - 398773), lithium-postgres (#14 - 398258), lithium (#45 - 271989). The last result is MySQL.
Further, for every framework/language that was tested with both MySQL and postgres (there's quite a few of them), the postgres one always ranks higher.
Perhaps it's because pg had better async drivers?
This is the "workload".
> [Multiple Queries] This is the first test where Just(js) has quite a big lead. This is likely due to the fact it is using a custom postgres client written in Javascript and taking full advantage of pipelining of requests. It also avoids sending a Sync/Commit on every query. As far as I am aware this is within the rules but will be happy to make changes to sync on every query if it is not.
https://just.billywhizz.io/blog/on-javascript-performance-01...