Comparing query rates (was Re: 9.2.5 db causes high cpu?)

[Jim Reid]

>>>>> "Brad" == Brad Knowles <brad at stop.mail-abuse.org> writes:

    Brad> 	For serving the root zone, after making sure that the
    Brad> cache was fully primed, I found that BIND-8 could handle
    Brad> ~490 queries per second, while BIND-9 did ~318.  In
    Brad> contrast, CNS did 777 qps.  For serving the .tv zone, after
    Brad> making sure it was fully cached, I found that BIND-8 did ~91
    Brad> qps, BIND-9 did ~55, and CNS did ~110.

Brad, please be more careful before publishing these results. Without
providing the context of your test environment, the numbers are quite
misleading. IIRC the results you got were from very modest hardware
resources that would not have been representative of the platforms
used for real DNS service since the early 1990s. I'm not disputing
your numbers: I am disputing their relevance. ISTR you saying in your
LISA talk that nobody should pay attention to the absolute numbers and
the focus should be on the comparitive results. That said, I wonder if
the limited memory that your test server had may have meant some name
servers thrashed, adversely affecting their throughout. Even for
comparitive purposes.

FYI, I just did a crude test on an elderly 300 Mhz Pentium II running
BIND 9.3. It handles ~2000 qps -- 20 times what your test found --
when answering for in-core authoritative data.