Re: dnscache performance still applicable today?

Matthew Dempsky wrote:
> On Sun, May 24, 2009 at 6:50 AM, Rihad <[email protected]> wrote:
>> Am I right in assuming that dnscache's single thread of execution becomes
>> its main bottleneck, so it has to delay query execution while processing
>> current requests?
> 
> The evidence you presented does not imply that.
> 
> You should look at dnscache's CPU usage instead.  If it's under 100%,
> then making it multithreaded will not help.
> 

   PID USERNAME    THR PRI NICE   SIZE    RES STATE  C   TIME   WCPU COMMAND
45536 Gdnscache     1  48    0   481M   479M select 0  24:00  6.15% dnscache

multiply 6.15% by 8 (because 2 quad-core Xeons are at work), still 
getting only about 50% CPU load. I wasn't expecting CPU to be the 
bottleneck anyway. It's the "receive request/fetch data/send response" 
cycle that blocks, because network roundtrip is centuries compared to 
cpu cycles. I don't think dnscache can do multiple requests concurrently 
under current design.