tuning routing using cxgbe and T580-CR cards?

[Olivier Cochard-Labbé]

On Fri, Jul 11, 2014 at 8:03 PM, Bjoern A. Zeeb <
bzeeb-lists at lists.zabbadoz.net> wrote:

> On 11 Jul 2014, at 17:28 , John Jasem <jjasen at gmail.com> wrote:
>
> > c) the defaults for the cxgbe driver appear to be 8 rx queues, and N tx
> > queues, with N being the number of CPUs detected. For a system running
> > multiple cards, routing or firewalling, does this make sense, or would
> > balancing tx and rx be more ideal? And would reducing queues per card
> > based on NUMBER-CPUS and NUM-CHELSIO-PORTS make sense at all?
> > ...
> > g) Are there other settings I should be looking at, that may squeeze out
> > a few more packets?
>
> If you are primarily forwarding packets (you say "routing" multiple times)
> the first thing you should do is turn off LRO and TSO on all ports.
>

Hi Bjoern,

I was not aware of disabling LRO+TSO for forwarding packet.
If I read correctly the wikipedia page of LRO[1]: Disabling LRO is not a
concern of performance but only of not breaking the end-to-end principle,
right ?
But regarding TSO[2]: It should improve performance only between the TCP
and IP layer. But paquet forwarded didn't have to cross TCP<->IP layer,
then disabling TSO should not impact performance, right ?

I've tried to benchs the differences on my lab:
- Hardware: quad cores (Intel Xeon L5630 2.13GHz, hyper-threading disabled)
with 2 ports Intel 10-Gigabit X540-AT2
- Multi-flows (different UDP ports) of small packet (60B) at about 10Mpps
(pkt-gen -f tx -i ix0 -n 1000000000 -l 60 -d 9.3.3.1:2000-9.3.3.1:4000 -D
a0:36:9f:1e:28:14 -s 8.3.3.1 -w 4)
- Result collected on the receiver side in Paquet-Per-Second unit.

ministat -w 74 tso.lro.enabled tso.lro.disabled
x tso.lro.enabled
+ tso.lro.disabled
+--------------------------------------------------------------------------+
|   +  +     x+    *                          x+                    x     x|
||____________M_|_A________________|________A_M_________________________|  |
+--------------------------------------------------------------------------+
    N           Min           Max        Median           Avg        Stddev
x   5       1724046       1860817       1798145       1793343     61865.164
+   5       1702496       1798998       1725396     1734863.2     38178.905
No difference proven at 95.0% confidence

=> There is not difference: Then I can disable LRO for respecting the
end-to-end principle. But why disabling TSO ?

Regards,

Olivier

[1] http://en.wikipedia.org/wiki/Large_receive_offload
[2] http://en.wikipedia.org/wiki/Large_segment_offload