Re: Route Information Options in Redirect Messages (updated)

[james woodyatt <[e-mail]>]

Hi Lorenzo,

I’m writing here to concur explicitly with everything Fred wrote in a his recent reply to this, and also to add my response to a particular point.

On Feb 6, 2017, at 11:51, Templin, Fred L <[e-mail]> wrote:
> On Feb 5, 2017, at 17:26, Lorenzo Colitti <[e-mail]> wrote:

>>  
>> More in general it looks like this document is trying to reinvent a fair amount of stuff that's already covered in detail, and more robustly, in homenet. Why not use those solutions instead?
> 
> […] About reinventing homenet, I think you would find that these concepts have been around a lot longer than home net.

Also, I’m not sure to what you’re referring here, Lorenzo. From my perspective, this draft is meant to address a problem that HOMENET has not yet addressed in its long history. In fact, I wrote to HOMENET on 19 Jan asking for review and comment on this draft. (Thanks for responding!)

The question I asked HOMENET is one this draft tries to answer: in the real world, how do routers inform *hosts* (not just other HOMENET routers) about optimal paths to more-specific routes?

The question is relevant, because in the real world, typical general-purpose home devices do not have RFC 4191 Type C behavior, and they will likely never have Type C behavior, for all the reasons that major operating systems vendors generally do not feature it in their factory default configuration, and most don’t even surface a reasonable human interface for motivated users to enable it. Because RA validation is difficult and RIO is a big big big rock to throw at a Type C host.

Clearly, relying on hosts to have RFC 4191 Type C behavior is insufficient. Hence the need for this draft. What in particular do you think HOMENET did here that is better? If I overlooked something, then I’d be grateful for any pointer you could provide.


--james woodyatt <[e-mail] <mailto:[e-mail]>>