Re: em forwarding performance (was Proposed 6.2 em RELEASE patch

At 12:57 PM 11/30/2006, Ivan Voras wrote:
Mike Tancsa wrote:

> Yeah I inadvertently slighted the NetBSD folks by leaving them out. So
> I guess I better give them a try as well.
>
> The part that really surprises me is the drop in performance as firewall
> rules are added to RELENG_6 and above. Both LINUX and RELENG_4 seem to
> scale well with the number of rules added but RELENG_6 takes a big drop.

Wasn't there some important setting in ipfw you can tweak if you need
lots of ipfw rules? Size of some hash table?

Quick Googling found this: http://info.iet.unipi.it/~luigi/ip_dummynet/
and net.inet.ip.fw.dyn_buckets: 256. AFAIK the hash size needed to be
tweaked manually in the code, and net.inet.ip.fw.dyn_buckets: 256 is
listed as read-only so this might be it. Maybe mailing Luigi will help
finding out...

I was told offlist "there is additional per-packet locking overhead not seen in RELENG_4 where all processing is covered by the same spl."...

---Mike

_______________________________________________
freebsd-performance@xxxxxxxxxxx mailing list
http://lists.freebsd.org/mailman/listinfo/freebsd-performance
To unsubscribe, send any mail to "freebsd-performance-unsubscribe@xxxxxxxxxxx"