Re: new to cdn internetworking

[Christian Kuhtz <ck@arch.bellsouth.net>]

On Tue, Nov 13, 2001 at 02:24:27PM -0500, Eric Dean wrote:
> 
> I don't disagree, but arguably 50% of all Internet requests are proxied.
> I'm not a proponent just stating the facts.

Ok, you got my attention, do you mind revealing your source for 
these facts?

Whether I'm a proponent or not has nothing to do with whether or
not an approach is ultimately viable (it simply changes the bias).

> It is also arguable which direction service providers are heading.

I'm not sure how it is arguable when one has first hand knowledge 
of such efforts as it is the case for me and several others around
the industry.

What evidence do you have that providers are still continue to 
embrace and grow their footprint in the direction of increasing
inband proxy based services?  (Leaving aside the whole discussion
of whether they scale in the first place; let's just assume they
do).

> I agree that backbone networks are certainly not using proxies; 

I apologize if it came across as being purely aimed at backbone
networks, that wasn't at all the intended point.  

This applies just as well for the edge services -- proxies in the 
core, as you probably well know, haven't been viable (if ever) 
for a long time.

> however, there are a variety of subscription-based services using
> technologies such as L2TP,IPSec, PPOE, PPOA that are overlays that
> offer things like content filtering and other value-adds.

Care to offer a few examples of working (read: fully
operationalized), financially viable models?

I have yet to see anyone whose 'content service and other 
value-adds' is grounded in reality.  Might you be the first?

I think the foundation for this math, the assumptions which
make this model viable, are rapidly changing.  And that is what
I was trying to point out.  They're changing in favor of not
having an existing base of proxy caches.

More specifically the question would be.. Are you saying you can
make a profit in the real world on the capital expense for
hardware, software licenses, support costs and operational costs
by deploying entirely new equipment for the sole purpose of
content overlays?

In the past several years, all the big guys have presumed that
when it came ot content distribution there was already a cache
type infrastructure there and this was an add-on service.  
Therefore, the only math one had to do is that one could offset
the additional cost of hardware & software & support &
operationalization by additional income (since the foundation
was assumed to be profitable (or at least break even when
used to provide the additional income) in the first place.)

That alone was hard to prove.  Not to mention that the whole
math of forward proxy caching ROI was flakey to begin with.

Show me somebody who is more than a niche player and is happy
with their forward proxy style caching or content solution.
'Happy' is defined as having ROI better than 0.

> I do agree that caching proxies are not being deployed inline
> with giga-routers though.

This discussion stretches all the way down to individual PPPoX
termination edge boxes.  And the whole point was directed at 
any sort of forced inband activity.

My point is that even for edge services (such as DSL etc) forward
proxies are actively being shut off/retired/trashed/scrapped or
preparations for such moves are in the works/have been for some
time.  A lot of people are looking at the financials and just
can't see them work (anymore, if ever).

Ask those people who still have them in their network whether
they are truely achieving any of what they were intended to do:
reduce the impact of another cost center.

The reasons for abandoning the path range from operational
nightmares to reality setting in on ROI myths to many others.

This is a reality today at several very large providers of such
services, but anyone is welcome to chose to ignore it. :)

Some of the very same people/companies working on CDN 
infrastructures in this forum are affected by exactly all that.
So, it seems little surprising that there's a bunch of people
that favor this.

I think for the sake of all of us, I think we're better off
without overlay networks and without forced inband manipulation.
I think I already argued why in this and the previous email.

Instead of further enhancing a broken model, a step beyond what
we have today is needed, not just from the point of distribution,
clearinghouse etc abilities of content networks.

Don't you think you get a much more financially and
architecturally viable solution by not persuing an overlay 
approach?

We can all decide to live in an architectural ivory tower, but
we also have to recognize some of the harsher biz realities if
we want all or any of this to fly.

You're welcome to try to convince me otherwise.  I'm listening
and all ears, and will always entertain fresh new ideas for old
problems.  

And if there's something that can be made work financially, I'm 
sure we're all very ready to hear about it.

If there's another service provider here that disagrees with me,
please speak up (publically or private).  I'd love to hear from
you.

Cheers,
Chris