[ovs-discuss] performance drop 60% when using vxlan tunnel

[Madhu Challa]

Like Jesse pointed out the numbers should be much better with offload. Here
are the numbers I see. My offload card is an eval card and is messing up
some packets. I am going to try it with latest firmware and update the
slides.

http://www.slideshare.net/MadhusudanChalla/ovs-perf

Thanks.

On Mon, Nov 17, 2014 at 9:18 AM, Rukhsana Ansari <
rukhsana.ansari at oneconvergence.com> wrote:

> Here are two old links that provide perf numbers with OVS GRE/VxLAN
> tunnels:
> http://networkheresy.com/2012/06/08/the-overhead-of-software-tunneling/
> http://networkstatic.net/configuring-vxlan-and-gre-tunnels-on-openvswitch/
>
> With br MTU= 8900 bytes,  the bare-metal OVS-vxlan numbers I see is 5.3
> Gbps with a single iperf thread and 9.3 Gbps with 2 iperf threads.
>
>
> -Rukhsana
>
> On Thu, Nov 13, 2014 at 9:04 AM, Rukhsana Ansari <
> rukhsana.ansari at oneconvergence.com> wrote:
>
>> Hi
>>
>> On Thu, Nov 6, 2014 at 7:27 AM, Jesse Gross <jesse at nicira.com> wrote:
>>
>>> On Wed, Nov 5, 2014 at 10:58 PM, FengYu LeiDian
>>> <fengyuleidian0615 at gmail.com> wrote:
>>> > 于 2014年11月06日 00:08, Jesse Gross 写道:
>>> >
>>> >> On Tue, Nov 4, 2014 at 11:03 PM, FengYu LeiDian
>>> >> <fengyuleidian0615 at gmail.com> wrote:
>>> >>>
>>> >>> Hi
>>> >>>
>>> >>> Env: redhat 6.4, OpenvSwitch-2.1.2, using native
>>> >>> datapath/linux/openvswitch.ko module
>>> >>>
>>> >>> VM1 on host1, VM2 on host2, host1 and host2 are connected by a
>>> switch.
>>> >>> both VM enable virtio/vhost when launching VM
>>> >>>
>>> >>> case1:
>>> >>> VM -> tap -> ovs-bridge -> eth1
>>> >>>
>>> >>> case2:
>>> >>> VM -> tap -> ovs-bridge -> vxlan -> eth1
>>> >>>
>>> >>> When using vxlan in case2, iperf performance drop 60%,
>>> >>
>>> >>
>>> >> Your NIC probably doesn't support offloads (checksum, TSO, etc.) in
>>> >> the presence of VXLAN.
>>> >
>>> >
>>> > This is the default features supported by my NIC 82599 both in case1
>>> and
>>> > case2.
>>>
>>> I can assure you that this NIC does not support VXLAN and the stack is
>>> being forced to do segmentation in software.
>>>
>>>
>> I have seen similar numbers (2.3 Gbps), with bare-metal Linux and OVS 2.0
>> with vxlan encap -
>> with Intel 82599 on Ubuntu 14.04.
>> MTU of bridge with vxlan port was set to 1400 bytes.
>> Do we have reference data that  2+ Gbps number is the expected perf for a
>> NIC that does not support VxLAN offloads?
>> This older link shows 4+ Gbps with 1500 byte MTU for bare-metal vxlan
>> perf with OVS (1.8.9):
>> http://networkstatic.net/configuring-vxlan-and-gre-tunnels-on-openvswitch/
>>
>> Thanks
>>
>
>
> _______________________________________________
> discuss mailing list
> discuss at openvswitch.org
> http://openvswitch.org/mailman/listinfo/discuss
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://openvswitch.org/pipermail/ovs-discuss/attachments/20141118/82e05748/attachment-0002.html>