Subject: Re: named hostname-lookup rate on NetBSD?
To: Michael Richardson <firstname.lastname@example.org>
From: Jonathan Stone <jonathan@DSG.Stanford.EDU>
Date: 04/10/2000 09:06:22
In message <200004101532.LAA00694@pzero.sandelman.ottawa.on.ca>Michael Richards
> How many names do you expect to cache?
"None at all". I'm looking up lists of, oh, half a million unique
hostnames. (I cant make the list available).
If/when I get it to go faster, I'll look up more ;).
> How much ram do you have?
>Are you swapping?
Nope. the named process doesnt get much over 60 megabytes. The CPU
load is about 70-75% in named. System time is only around 10-15%.
(Again, this is on FreeBSD. I'll have to juggle some hardware to try
it on NetBSD.) The machine is otherwise idle, except for the 500 or
so processes iterating over gethostbyname().
> As named is not (to my knowledge) multithreaded yet, I guess the second CPU
>capability of FreeBSD won't help much.
BIND 9 uses pthreads, but I havent looked closely enough to see if
BIND 9 would ever acutally benefit from two CPUs, let alone in this case.
> Jonathan> Anyone know how NetBSD (1.4.2 or -current) would do? Same,
> Jonathan> better, or worse?
> I would imagine about the same.
Darn. Guess I'll have to try it and see, then fix the performance
I think it *is* a bug, since the time is spent in userspace, and it's
not rocket science to build, say, a webserver that can sustain
100Mbit/sec for static content on similar hardware.
(See, for example, Vivek Pai's "flash" server.)
I just thought someone would have statistics on NetBSD as a
heavily-used (thrashed) nameserver.