On Wed, 23 Sep 2026 12:44:27 -0400 Willem de Bruijn wrote:
> Wang Zhan wrote:
>> On Mon, 21 Sep 2026 17:07:09 -0400 Willem de Bruijn wrote:
>> > Somewhat tangential, but an entirely different practical approach
>> > could be to disable BIG-TCP when such a patch from BIG-TCP capable to
>> > non-capable device is encountered.
>> 
>> That means a single device in the topology which does not support BIG TCP
>> forces every other device to turn BIG TCP off, to avoid the degradation.
>> So with mixed devices and a dynamic topology, BIG TCP cannot be enabled at 
>> all.
> 
> True. A question is how likely such a scenario with mixed devices is.
> I.e., is it worth the code complexity.

I cannot say how common the mixed case is, but with this patch enabling
BIG TCP on a device becomes a local decision: if the peer supports it the
path is faster, and if it does not the path stays at what it would be
without BIG TCP.
_______________________________________________
dev mailing list
[email protected]
https://mail.openvswitch.org/mailman/listinfo/ovs-dev

Reply via email to