BIND 9.x caching Performance under heavy loads

[Rick Jones]

Srini Avirneni <avirsri at gmail.com> wrote:
> Yes, it is a suprise. Its a surprise to see Bind as a process, peak
> at 25% CPU day 1. Then, Day 2, peak climbs to over 50%. Load has not
> changed.

> By day 3, CPU will exceed 75% and Bind will no longer respond in
> a meaning full way (< 500 queries/sec). This is on very fast
> hardware.

> This was with peak query rates of 2500/sec ranging down to 1200/sec.

> As load increased, the CPU time ran away quicker. One would expect
> that a Bind server that runs at 25% CPU during peak, would within a
> few points, operate as such continually, and scale somewhat
> linearly.

Clearly we need to find some way to gather profiles of the process
each day to see which routines start consuming more time.  If this
were PA-RISC or Itanium I'd suggest either Prospect for PA-RISC
(http://www.hp.com/go/prospect) or Caliper for Itanium
(http://www.hp.com/go/caliper) or perhaps q-syscollect for Itanium 2.6
linux (http://www.hpl.hp.com/research/linux/q-tools/)

The important thing is to find a tool for your platform that can
attach to a running process and get the profile - clearly having to
stop and restart the process is going to alther the behaviour, and I
we cannot just start it profiled and leave it that way for N days -
too much chance of counter overflow and the effect of previous days
may not be fully overcome.

rick jones
-- 
portable adj, code that compiles under more than one compiler
these opinions are mine, all mine; HP might not want them anyway... :)
feel free to post, OR email to raj in cup.hp.com  but NOT BOTH...