Well I must admit that you, Ian, attend all the meetings
and thus can express your thoughts to all.
X-Mailer: QUALCOMM Windows Eudora Version 5.1
Date: Tue, 19 Jun 2001 10:24:50 -0700
To: "Michael W. Condry" <condry(_at_)intel(_dot_)com>
From: Ian Cooper <icooper(_at_)equinix(_dot_)com>
Subject: Re: WG Review: Open Pluggable Edge Services (opes)
Cc: ietf(_at_)ietf(_dot_)org, ietf-openproxy(_at_)imc(_dot_)org
At 07:55 6/19/2001 -0700, Michael W. Condry wrote:
Our interest in OPES and the interest of the folks we are working with
are not with services such as unrequested ad insertion or other items
be viewed as offensive. Lots of things can be mis-used, SPAM email
is a better example; we even get it on the working group mailing lists, ever
read the WEBI mailing list?
I don't think this is a fair example. Unless you're comparing OPES
services with open-relay SPAM filters? (OK, sorry, that's probably
another Pandora's box I just opened on this list...)
Although you have made many comments on the charter as well
as your thoughts during the open area applications meeting I have not
seen you at the OPES BOFs or OPES workshops.
*cough* not everyone can make it to physical meetings, and while I suspect
that Keith (since that's who you are addressing) is physically present at
the IETF meetings it's very possible that he's actually sat in another
meeting that's take place at the same time.
That's why we have the mailing lists.
There you could
hear a diverse set of applications that I would be surprised if you would
feel to be offensive. We plan to have a '"Deployment Scenarios" document
that describes a range of applications and constructive comments from
you would be appreciated.
Offensive services can be done at the server end of things as well, such
as ad-insertion; so the edge architecture is not the issue. We have
recorded your key point that is doing modifications on the content
without the permission of the "ends" (client and server) of the operation.
OPES is quite clear about this requirement.
I don't think that's quite the objection, though of course I could be wrong.
I haven't seen anyone state *clearly* what the objection is to OPES-stuff
with relation to the end-to-end nature of the Internet (and why it's still
applicable even when you have an end-to-end connection between browser and
OPES proxy). So here's an attempt (with thanks to the work on the midtax
document for getting me thinking about this)...
The problem with OPES when deployed in the "general Internet" is that as
an end user I may experience substantially different experiences
(receiving very different content) depending on my entry point. So, for
example, I may suddenly find that I'm using a network that doesn't have
that virus scanning service I was relying on. Or the French-to-English
translation service I needed would suddenly be unavailable, so I couldn't
browse a favorite site any more.
And yes, general caching proxies cause problems too - I can experience
different views because of aggressive caching or the absence of caching.
That said, it's possible that as a user I could authenticate myself with
my home/usual Web intermediary so that I could still get at my
services. If they're available of course (what happens if I'm at home and
the OPES proxy goes down - I don't get my virus scanning any more do I...)
Michael W. Condry
Director, Network Edge Technology