A few comments on draft-day-cdnp-model-02.txt

["doug potter" <dougpott@cisco.com>]

Don, Gary, and Barron covered most of my comments already. Here's the rest:

Section 2.2 - I think we should remove the 2nd paragraph (2
    sentences). It too narrowly defines a CDN as "effectively 
    a collection of widely-dispersed caches". Paragraph 4 is
    a better overall description.

    Paragraph 4: Change "move a client toward a rendezvous with 
    a content server" to "move a client toward a rendezvous with
    a surrogate".

Section 3 - I agree with Barron that CONTENT and CONTENT DATA
    UNITS as defined is a little much. Discussing what is known
    and unknown by the CDN adds no value to this document.

    The definition of SURROGATE states "Receives a redirected
    REQUEST and...". The word 'redirected' should be removed.

Section 4.1 - The comparison between pre-CDN and CDN is lost on
    me. To me, CDN is a name for something everyone was already
    doing rather than a change in the way folks did things. The
    simplest CDN is a DIRECTION SYSTEM (DNS server) and an ORIGIN
    SERVER. Certainly any multi-site set of ORIGIN servers is
    also a CDN. It has a DIRECTION SYSTEM and selects sites based
    on some metric. At a minimum I'd remove the first two
    paragraphs. But the third paragraph repeats the last paragraph
    of 2.2. I'm inclined to remove the whole section.

Section 4.2 is also pretty obtuse. Is anything of value lost if we
    remove the first paragraph? 

Section 5 - I think AUTHORITATIVE DIRECTION SYSTEM belongs in
    section 3. It is not specific to the Peering Model. 

    DIECTION SYSTEM is described in section 3. And DIRECTION
    PEERING is in section 5. DIRECTION PEERING SYSTEM is not
    defined anywhere (but is used in the DIRECTION ADVERTISEMENT
    definition).

drp