[ovs-discuss] [openflow-discuss] Scalability testing
blp at nicira.com
Sun May 29 20:37:03 UTC 2011
On Sun, May 29, 2011 at 09:34:13PM +0200, Jose Luis Franco Arza wrote:
> May I ask you for a little bit of help? I'm doing my master thesis related
> to a scalability test of the OpenFlow protocol. I was looking for a limit or
> bottleneck in the NOX controller but after changing the line you told me the
> behaviour of the NOX controller was perfect.Now it results that the
> bottleneck is in the openvswitch. It can't handle more than 2000 flows/sec
> because the CPU processing increases a lot when trying to forward packets
> with a different UDP port each one.
> I have tried to ask in the openvswitch discuss forum but I didn't have any
> http://openvswitch.org/pipermail/discuss/2011-May/005236.html (There is a
> file attached with the results of the experiments).
You asked a question on Saturday (of a 3-day weekend in the U.S.) and
complain on Sunday that you didn't get a response? Be realistic.
More information about the discuss