[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [DNA] Definition of "Link Up" and "Link Down" events?



> With the above definitions we'll minimize what DNA would consider
> spurious notifications just because of signal strength, yet never miss
> the case when the layer 3 link is different.

Yes, I think that is a reasonable approach.

> When a host has multiple NICs there might be any number of policy
> related ways the host wants to choose.

I think it's important to distinguish what choices we are talking about.
If we are talking about choosing which NIC to transmit on, that is
influenced by a variety of factors such as routing metrics, Strong/Weak
host model.  So link indications only influence this to the extent that
they are taken into account within the routing metrics.  Just because a
link is not preferred in terms of routing to a particular destination
prefix does not mean it is "down".

If we are talking about the decision to bring a link up or down (e.g.
"dial in demand" or "idle timeout") that is a different decision.  I'd
argue that "Link Up" or "Link Down" events are the consequence of these
decisions.

> This could be based on a combination of signal strength, "flakiness
> history", cost, and it might be different for different applications
> running on the same host.

"Cost" and "flakiness history" (frame loss) have in the past been
incorporated into routing metrics, whereas signal strength has been used
in the decision about whether to bring a link up, since in these
circumstances frame loss cannot yet be observed.

In most implementations all applications share the same routing
table and interfaces, and the decision to bring links "up" or "down" is
typically handled below the application, which makes the decision on which
interface(s) to bind to.

> saying that the host implementation can benefit from finding out about
> significant changes in signal strength, observed L2 retransmissions,
> etc., as some of the inputs to the hosts policy choices.

Sure, but I'd suggest that we should be clear that this additional info is
not a determinant of "Link Up" and "Link Down" events.