On Mon, Jul 31, 2000 at 04:37:32PM -0600, Hilarie Orman wrote:
> Thanks for the comments. On some of them, I'm not sure what you want
> ... background of the parties? Raised in Fresno, California and currently
> living in Utah? Novell is a company headquartered in Provo, Utah and
> sells a caching product? Sun is a big company?
Sorry, the question was poorly put. I'm interested in the context of the
project; your web page refers to an 'EPSFW charter', which implies a few
things; I'll wait for your presentation.
> We tried to start with requirements for an extensible services
> environment, prior to having a full architecture. We do want to
> standardize the ruleset matching (something that the IETF Policy WG is
> touching on these days) and proxylets.
Great. I'm excited by what you're doing, because to me it represents an
opportunity to develop a standard set of requirements and a framework in an
umbrella over proxies, surrogates, CDNs using surrogates, etc. and has the
potential to address many of the areas that I looked at in my surrogates
paper, for all of these devices.
> Sorry we failed on terminology. The term "surrogate" was being defined as
> we were writing (and I tried to help with the definition). We like to be
> tidy wrt to the WREC usage, so if you could give a couple of misuse
> examples we could clean them up pretty quickly.
The most noticable is 'proxy', when many of the uses described are in a
surrogate role, which is separate. Surrogates are gateways, which is a
different kind of intermediate in RFC2616.
So, it would be best to call this the 'Extensible Intermediate Services
Framework', and so on.
> It's possible that entities "in flight" don't need entity names. However,
> the entire area of variant naming needs examination. An ephemeral object
> isn't necessarily a variant, though.
There should probably be requirements re: entity identity if the entity is
changed; it should either be stripped or changed, but there are probably
other issues lurking in there as well.
Varience is very interesting in the context of an extended service
intermediate.
> And I'm not sure what you mean by "synthetic" objects (those that don't
> exist). Perhaps the roadmap should include a terminology document.
Using ICAP as an example, Request Satisfaction mode allows construction of
objects at the 'edge', effectively making the intermediate an end point,
whether it's a proxy or a surrogate. Distinguishing this object from one
which *would* have been returned had ICAP not taken place can be
problematic, for purposes of configuration, metadata, logging, etc.
> It's surprisingly difficult to find people here in Pittsburgh through
> Brownian motion, but I'd be glad to talk about these issues in person if
> you collar me; otherwise, look at the message board for Thursday evening
> plans.
I'm coming in on Thurday night; will try and catch you there, or on Friday.
Cheers,
-- Mark Nottingham, Research Scientist Akamai Technologies (San Mateo, CA)
This archive was generated by hypermail 2b29 : Thu Nov 18 2004 - 11:21:29 MST