The results you are referring to were collected a year and a half ago
and the code base has evolved significantly since (including many
performance optimizations). Open vSwitch makes aggressive use of
flow-caching so the addition of new features should not affect packet
forwarding rates in the common case.
These are nice results. It would be great if you could keep us updated
as your research progresses.
.martin
Hi.
During my research I did some performance tests. In some of your paper
on ovs I read that "Open vSwitch is almost as fast as the native
bridge". Well in my environment ovs beats the Linux Bridge [1].
Do you have any explanation why Open vSwitch, with all it features, is
faster than the "simple" Bridge?
My setup is:
- Host:
o kernel 2.6.36.1
o ovs 1.1.0pre2
o one bridge
- 2 guests:
o kernel 2.6.36.2
o QEMU 0.12.4
o TAP interface (attached to bridge)
- measurement
o iperf
o 15 runs
o 5 minutes each
Thanks,
Daniel
P.S.
Don't get me wrong, I am not complaining ;-)
[1] http://i.imgur.com/KgH0J.png
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss_openvswitch.org
--
~~~~~~~~~~~~~~~~~~~~~~~~~~~
Martin Casado
Nicira Networks, Inc.
www.nicira.com | www.openvswitch.org
cell: 650-776-1457
~~~~~~~~~~~~~~~~~~~~~~~~~~~
_______________________________________________
discuss mailing list
[email protected]
http://openvswitch.org/mailman/listinfo/discuss_openvswitch.org
