link monitoring and BFD in SDN networks

Dave Bell me at geordish.org
Mon Jan 19 22:55:04 UTC 2015


BFD etc aim to prove there is end-to-end connectivity between two
points, not just that all links are up along the path. All ports could
be up, but end-to-end connectivity broken, for example a misconfigured
VLAN across a L2 network. Sending some kind of packet across the
network is pretty much the only way to guarantee reachability.

The OpenFlow protocol in particular has a way to instruct a switch to
send a frame out of an interface. By default, the OpenFlow switches
will forward all frames it has received and doesn't know what to do
with back to the controller. This means someone could write an OAM
protocol that will work via OpenFlow. A quick google for 'OpenFlow
OAM' brought me this link which has someone who has done just that:
"http://www.rvdp.org/presentations/SC11-SRS-8021ag.pdf"

Of course if you want fast failover, you need to send packets very
rapidly. Every 250ms is not unreasonable. This is going to cause the
control plane to get very chatty. Typically on high end routers,
processes such as BFD are actually ran on line cards as opposed to on
the routing engine. When a failure is detected this reports up into
the control plane to trigger a reconvergence event. I see no reason
why this couldn't occur using SDN.

Regards,
Dave

On 19 January 2015 at 22:01, Glen Kent <glen.kent at gmail.com> wrote:
> Hi,
>
> Routers connected back to back often rely on BFD for link failures. Its
> certainly possible that there is a switch between two routers and hence a
> link down event on one side is not visible to the other side. So, you run
> some sort of an OAM protocol on the two routers so that they can detect
> link flaps/failures.
>
> How will this happen in SDN networks where there is no control plane on the
> routers. Will the routers be sending a state of all their links to a
> central controller who will then detect that a link has gone down. This
> just doesnt sound good. I am presuming that some sort of "control plane"
> will always be required.
>
> Any pointers here?
>
> Is there any other reason other than link events for which we would need a
> control plane on the routers in SDN?
>
> Thanks,
> Glen



More information about the NANOG mailing list