|
Fabrice wrote:
>
> Hello Joe,
>
> That 8000 packet/s limit doesn't exist, look on the mailing list for
> threads "SYN floods and LVS-NAT CPU Load" and "LVS and
> conntrack".
What's your explanation then for the experimentally observed fact that
the packet rate for a tcp connection is constant for all packet sizes?
See the first paragraph in
http://www.linuxvirtualserver.org/Joseph.Mack/performance/single_realserver_performance.html#netpipe-mtu
and the graph just below it.
> With my 100 Mbps network, I was able to send upto
> 60'000 SYN/s. Ok they aren't "standard" TCP packets but it still
> proves that you aren't limited to 8000 packets/s.
something (I don't know what) is different about SYN packets.
> You only have a 8000 packets/s limit when you send full MTU
> packets.
not on my setup.
> Do you have connection tracking modules/code loaded in your
> kernel?
no
Joe
--
Joseph Mack PhD, Senior Systems Engineer, Lockheed Martin
contractor to the National Environmental Supercomputer Center,
mailto:mack.joseph@xxxxxxx ph# 919-541-0007, RTP, NC, USA
|
