Maintained by: NLnet Labs

[Unbound-users] Unbound multithread performance: an investigation into scaling of cache response qps

W.C.A. Wijngaards
Tue Mar 23 15:28:35 CET 2010


-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hi Simon,

On 03/23/2010 03:20 PM, Simon Perreault wrote:
> On 2010-03-23 01:38, W.C.A. Wijngaards wrote:
>> Configuration ------- 1 core --- 2 cores --- 3 cores --- 4 cores
>> select and pthreads     8450      14100       16100       18600
>> select and solaristhr   8600      13800       15800       17500
>> select and no threads  10000      17800       19800       22800
>> evport and pthreads     8400      13600       15900       18100
>> evport and solaristhr   8500      14100       16000       18600
>> evport and no threads   9700      17300       19600       22300
> 
> Very interesting! Thanks for sharing this.
> 
> I have one question: why does "no threads" go faster as the number of
> cores increase?

Without threads unbound creates processes (with fork(2)).  Every process
is a complete resolver, the interprocess communication is done with
pipes in the background (which have no traffic until you send an
unbound-control command).

Best regards,
   Wouter
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Fedora - http://enigmail.mozdev.org/

iEYEARECAAYFAkuo0BIACgkQkDLqNwOhpPgA5wCglryh80uMocI86HIYmkVh6lLx
O8sAn1htmu0dKNTdV+qige2FLYQ7vnJI
=U8K+
-----END PGP SIGNATURE-----