[dns-operations] DNS server benchmarking sanity check

[Shane Kerr]

Hello,

On August 15, 2016 11:29:32 AM GMT+08:00, Mark Delany <x2t at foxtrot.emu.st> wrote:
>On 14Aug16, Robert Edmonds allegedly wrote:
>
>> Did you happen to test a unique port vs. a unique IP address? I.e.,
>it
>> sounds like you tested *:53, *:54, *:55, etc. What about
>192.0.2.1:53,
>> 192.0.2.2:53, 192.0.2.3:53, etc.?
>
>Sadly no, I did not test that.
>
>Heck Robert, why didn't you suggest this two years ago when I was in
>the thick of it? In retrospect, it's such a beneficial and obvious
>thing to try. Drat.
>
>My total guess is that the kernels are doing some sort of hash onto a
>lock table based on the destIP, destPort tuple so a unique destIP
>would be as useful a tie-breaker as destPort appears to be.
>
>Having said all that, many of these issues seem obvious. Is it really
>the case that UDP kernel performance has not been extensively analyzed
>in the context of DNS servers?

The Linux kernel has had periodic regressions on much more performance-sensitive and mainstream loads like databases, so I doubt that anyone in the kernel space spends much time thinking about DNS if they don't care about SQL.

Cloudflare is possibly ahead of the curve here, and they blog about such things....

-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.