[ovs-dev] Issues with openflow protocol WAS(RE: [GIT PULL v2] Open vSwitch

Justin Pettit jpettit at nicira.com
Mon Nov 28 18:44:48 UTC 2011


On Nov 28, 2011, at 6:07 AM, Jamal Hadi Salim wrote:

> On Mon, 2011-11-28 at 13:54 +0000, Fischer, Anna wrote:
> 
>> Yes, I mentioned this months ago, and I am surprised this critical 
>> issue has never been picked up on and addressed. With a flaw like 
>> this there is no chance this component can be used in any serious 
>> virtualization deployment where different customers share the same physical server.
>> 
>> The path up to user-space needs to be designed in a multi-queue fashion, so that 
>> each vPort has its own queue up to user-space. Ideally those queues also need to 
>> be rate controlled in some form, so that no DoS is possible.
> 
> Good - more folks scrutinizing openflow ;->

I realize you chair an IETF standard with overlapping goals with
OpenFlow (ForCES), so you may have strong opinions about its design.
However, that's not relevant to this discussion, since OpenFlow's design
has nothing to do with the discussion being held here in regards to Open
vSwitch.  OpenFlow is just a bullet point--although an important one--in
a large set of features that Open vSwitch provides.  Its design is such
that it should be fairly easy to include new control protocols; OpenFlow
is just a library in Open vSwitch.  If you have issues with OpenFlow,
those would be better directed to the ONF or one of the OpenFlow mailing
lists.

--Justin

> That would resolve the kernel->user if in addition you then add
> prioritization of those queues. 
> But even then also the same problem exists with open flow in the 
> northbound direction towards the external controller where
> there is a single (TCP) socket.
> 
> cheers,
> jamal
> 
> _______________________________________________
> dev mailing list
> dev at openvswitch.org
> http://openvswitch.org/mailman/listinfo/dev




More information about the dev mailing list