On 19/04/15 11:55, Andrew Stuart wrote:
I ran the Nginx and benchmarked it using siege.
The performance was about 500 requests/second when I would have expected in the
range of 5,000 to 11,000 request per second for nginx on the machine that I ran
it on.
Did you obtain that reference number by running the benchmarking tool
and the httpd in different domains? I'd expect things to still be a lot
slower, but not by a factor of 10.
I’m currently doing more thorough and scientific testing to quantify the
performance issue but I thought I’d ask does this make sense?
I wouldn't bother benchmarking too much except if the interest is to
pinpoint and fix bottlenecks. There are still many known issues. For
example, the NIC layer needs quite a bit of love. It's still in the
same state performance-wise as when I originally made it work just to
get packets flowing.
The good news is that no socket errors were reported.
So you've been able to repeat mato's experiment where you take an
existing, real-world httpd, compile it as a rumprun unikernel, throw as
much traffic at it as you like, and it keeps on working. I think it's a
pretty significant result. Of course there's still a lot of work ahead,
but at least part one of "make it work, then make it fast" is there.
