Re: A DNS Based Mapping Peering System for Peering CDNs

["Abbie Barbir" <abbieb@nortelnetworks.com>]

I do not think that the two drafts should be collapsed into one. My view
is that there should be a running draft that describes  (or provides a
summary of ) various DNS based mapping methods and be used as a reference.

The draft should be updated on regular basis.

I do think that more methods will be proposed as the discussion on this
subject evolves.

However, the draft proposes a particular way for peering deliver aware
CDNs. I do like the work that was done by the authors to describe the
DNS-MS system in terms of trees. I think that the same result could also
be reached if a particular CDN is thought off as a collection of sub-CDNs
each specializing in the delivery of a certain type of content.

An important point that still to be discussed in the development of
method(s) that can determine the best surrogate based on its proximity to
the client. The challenge here is to get people to start hammering this
point.

Furthermore, the interplay between the distribution system and the
direction (or whatever the current name) is still need to be addressed.

------
Abbie Barbir, Ph.D.
Senior Designer
Nortel Networks
(613) 763 5229



Fred Douglis wrote:

> >Can you please explain how this compares to the "known mechanisms"
> >I-D previously submitted and discussed on this list?  Seems similar to
> >me at a quick glance.
>
> Actually, at a not-quite-so-quick glance, I saw that I-D referenced,
> and went back and searched for citations to it.  I see that there
> are a few comparisons scattered about that weren't immediately obvious,
> but I still wonder whether this addition isn't better suited as a
> revision to [5] rather than a separate draft.  Again, this is only on
> a fairly cursory examination, and by way of explanation.
>
> Fred