Re: new version of the reachability & exploration protocol

[Jari Arkko <jari.arkko@piuha.net>]

> Ah yes, but the diagram doesn't show that...

Ok, I fixed this for the next rev now.

> > Remember that this protocol
> > is designed to do both reachability and exploration at the
> > same time.
>
>
> I think it makes more sense to have different message types for this,  
> as the FBD keepalives will be relatively frequent but full  
> exploration hopefully much less frequent. Note though that I start a  
> full exploration after just missing one keepalive.

This isn't all that different from what I do... though we can
discuss the specific number or timing of what triggers exploration.

In my protocol 'I see you' == false means essentially that we
are still OK at least as far as the receiver's need for exploration
goes. So this is really just a keepalive for the receiver.

Conceptually, why I like a merged protocol is that it is
allows a smoother transition between reachability
and exploration. This can be important if you are still
hoping for a reachability test to succeed but already want
to proceed to trying out some other address pairs. Similarly,
this would seem to allow future extensions to load balancing,
exploration of alternative pairs while everything still works
etc.

> Well, the part that I wrote extensively about is very short in your  
> draft so it's hard for me to judge whether that part would be very  
> different.
>
> The most important difference that I see is that in my draft there is  
> per-context work and per-host work, while in your draft everything is  
> per-context.

I like your host/context difference approach.

We also need the other parts, including the nitty gritty details.
I'm hoping the two drafts together contain much of what is
needed.

> I'll be interested to see if my draft is accepted, as it doesn't have  
> any page breaks.

That's usually not a problem. The main problem they complain about
is the use of wrong legal copyright etc template.

--Jari