DNS performance testing - FreeBSD & Solaris - BIND & djbdns

[Kevin Darcy]

Brad Knowles wrote:

> At 7:25 PM -0400 6/21/01, Kevin Darcy wrote:
>
> >  Wait, you're comparing BIND's *forwarding* performance to its performance
> >  serving data from authoritative zones? That's apples and oranges. You should
> >  measure those separately.
>
>         That's why the tests are run multiple times -- the first time
> measures the forwarding performance, the second and third times
> measure the caching performance.

Well, I didn't see any reference to what the TTL settings were on the records, so
who knows much of the results are due to forwarding and how much to caching? If
the TTL setting were really low, you could have a mixture.

In any case, what I meant to question (poorly, it seems) was the apparent
assumption that BIND's performance answering from authoritative data is
automatically equivalent to its performance answering from ((forwarding) or
(caching) or (some combination of the two)). I don't think that's a safe
assumption.


- Kevin