Basic bind capacity question
Adam Tkac
atkac at redhat.com
Thu Jul 19 14:15:04 UTC 2007
Mike Toler napsal(a):
> Yes, I'm using a postgres DB backend. Using a single instance of the
> queryperf tool and a DB of ~120,000 entries, I'm getting 40 qps.
>
> Michael
>
If you query BIND with same RR (for example 50000 queries which asks to
RR which is stored in database) and you still have about 40qps it looks
you have misconfigured cache (your clients aren't allowed to access it)
so named have to search,search, search and this takes much time
(especially when you have 120,000 RRs). If you're using Dynamic Loadable
Zones (DLZ) postgres driver and you asks for different RRs it's possible
that this could completely degrades answer performance (because named
have to always asks postgres)
Adam
More information about the bind-users
mailing list