It takes CPU cycles to process the packet, but much fewer of them. The biggest
wins are that the packet doesn't need to be sent to userspace (and back) and
the kernel's exact-match hash lookup is much cheaper than the userspace's fully
wildcard-able classifier.
--Justin
On Jun 14, 2011, at 8
Justin,
Thanks again for your reply.
But i have then one more question: Flows installed in the kernell
doesn't consume CPU? Or is it that at the rates i'm using (50Mb/s),
the number of packets (100 byte-datagrams) aren't enough to consume
noticeable CPU?
I've noticed in the "ovs-dpctl show" that
On Jun 13, 2011, at 10:23 PM, Victor T. wrote:
> So i would like to ask now more specifically about the CPU usage of the
> OpenVSwitch and the different types of packet size in a flow.
>
> My test setup consists of 3 PCs, where 1 sends data to 3 by
> 2(OpenVSwitch,Nox). The traffic was generate
Hello,
I'm e-mailing this list as recommended by Justin in the OpenFlow mailing
lists.
So i would like to ask now more specifically about the CPU usage of the
OpenVSwitch and the different types of packet size in a flow.
My test setup consists of 3 PCs, where 1 sends data to 3 by
2(OpenVSw
