Comments on draft-day-cdnp-model-02.txt

[Gary Tomlinson <garyt@entera.com>]

Here are my comments from a close reread of draft-day-cdnp-model-02.txt.
I've made an attempt to take into consideration the comments made tonight by
Don Gilletti and Barron Housel and not duplicate them.

Gary

[review begins here]

Section 2

Para 1, 1st sentence,  in response to Barron Housel's comment on web
constrained model, I believe we have made a pragmatic decision to limit our
scope to the web (including streaming media) at this time.

Para 1, 2nd sentence, I believe the programmability inference pointed out by
Barron Housel  is better suited to the Open Extensible Proxy Services (OPES)
work, which can be overlayed onto CDNs.

General Problem.  The terms proxies, proxy caches, forward proxy caches,
reverse proxy caches, and caches are used pretty loosely.  All of these
terms were debated heavily a year ago within the WREC WG.  The resultant
Taxonomy doesn't define proxy cache (its essentially a slang term in the
industry) and defined cache to mean "a programs local store".  Also, since
we have defined SURROGATE more narrowly than the WREC Taxonomy, I suggest we
use the Taxonomy synomyms "server accelerator" and "reverse caching proxy"
in the appropriate contexts.  With this in mind, I've taken the liberty to
suggest the appropriate terms in each case.

Para 3, "use of server farms and proxies to improve", proxies by themselves
don't improve performance.  Substitute "proxies" with "server accelerators".

Section 2.1

Para 2,  3rd sentence, substitue"forward" with "caching proxies".
Para 2, 3rd sentence, substitute "reverse proxy caches" with "reverse
caching proxies".

Para 3, 1st sentence, substitute "reverse proxy caches" with "server
accelerators".
Para 3, 2nd sentence, substitute "reverse proxies" with "servers
accelerators".
Para 3, 3rd sentence, substitute "forward proxy caches" with "caching
proxies".

Section 2.2

Para 2, 1st sentence, substitute "caches" with "reverse caching proxies".
Para 2, 2nd sentence, substitute "caches" with "reverse caching proxies"
twice.
Para 3, 1st sentence, substitute "caches" with "server accelerators".

Section 3

CONTENT DATA UNIT, I concure with Barron Housel's comment on not wanting to
go into parsable content in the CDN.  This too IMHO is within the
application domain of the OPES work.

ORIGIN, 1st sentence, "BILLING CDN" with "CDN".

Section 5

AUTHORITATIVE DIRECTION SYSTEM, we aren't capturing the nuance here, of
authority of the object identifier name space being granted by the ORIGIN to
this direction system.  There are also implied semantics missing from the
definition pertaining to the root (aka FIRST DIRECTION SYSTEM) always being
authoritative and optionally delegating authority on a per request basis to
the federated peering direction systems involved in the recursive process of
directing a request to a surrogate..