[EME] RE: finally!: draft-irtf-eme-francis-nutss-design-00.txt

On Apr 24, 2007, at 11:10, Paul Francis wrote:

Folks,

> The main feature of the NUTSS is the "dual signaling" approach,  
> where both
> path-coupled and path-decoupled signaling are tightly coordinated  
> in order to
> overcome the limitations of either mode alone.  Some form of dual- 
> signaling
> seems necessary to me, but there might be many ways to manage it.   
> This draft
> suggests one way...others might have better ideas.

Here's another idea, this time a clean-slate one. While reading the  
draft, please

1) forget the data-orientation and concentrate to the policy aspects  
(in other words, we can name services/apps/hosts instead of data, if  
that's what apps desire),
2) replace an RH with P-box in your mind, and
3) do not let the different naming to distract you too much.

It strikes me that the similarities are very obvious:

(a) both do name based routing. NUTSS uses DNS, we do it with flat  
identifiers.
(b) both have two kinds of paths between end-points. NUTSS talks  
about off-path vs. on-path, we talk about slow-path and fast-path. In  
both, the policies are mostly applied on the off/slow-path, i.e., it  
is designed for *functionality*, not for speed like the off/fast-path  
is.

It seems that these two ideas are the in the deep core of both  
proposals. DONA is probably cleaner/simpler due to its clean- 
slateness, while NUTSS favors incremental deployability instead, and  
does that for a very good reason, of course.

Teemu

--

[EME] RE: finally!: draft-irtf-eme-francis-nutss-design-00.txt

Attachment: dona-final.pdf