Re: Section 7 of the architecture document

[Ian Cooper <icooper@equinix.com>]

At 15:24 11/3/00 -0500, Mark Day wrote:
>I don't really understand why section 7 is in the document.  I assume it has
>something to do with the various "interception proxy" traumas that took
>place in WREC, but I don't otherwise understand its role.  It's clear that
>(1) "end-to-end" is assumed to be good, and that
>(2) CDNs might somehow threaten "end-to-end", so
>(3) an argument is constructed about why this is actually not the case.
>
>But I'm not sure I believe any of 1, 2, or 3, at least as currently
>presented.
>
>What harm would befall us if we simply dropped section 7 entirely?

In itself I agree that the section doesn't appear to serve a particularly 
useful purpose.

However, while the "end-to-end" principle of IP is intact, the introduction 
of surrogates adds an interesting issue to the "end-to-end" nature of the 
transaction itself.  One would hope that the peering/invalidation/control 
mechanisms in place (and being developed) ensure that the surrogate's copy 
of a piece of content is identical to that on the origin server.  I.e. that 
the surrogate is semantically transparent.  That's something I think needs 
addressing.