RE: MPLS VPNs or not?

From: Fletcher E Kittredge [mailto:fkittred@gwi.net]
Sent: Wednesday, August 08, 2001 11:35 AM
To: Kavi, Prabhu
Cc: 'Vadim Antonov'; Christian Kuhtz; nanog@merit.edu
Subject: Re: MPLS VPNs or not?

> I guess the real question should be how much market cap did other
> companies lose because of certain people's zealotry? Any answers
> Vadim?

Prabhu;

  What evidence do you have that:

1) UUnet is/was a success,

My take is that the best measures of success is market acceptance
and profitability. Not a perfect measure, but certainly beats
alternatives like academic debates like protocol X is evil, or
service provider Y sucks. If protocol X is evil, ISPs will not
use it, or those that do use it will fail. If protocol X was
useful but has outlived its usefulness (e.g. ATM in the core),
ISPs will no longer use it. [Note that I am not against academic
discussions. However they are only useful until market deployment
has proven them right or wrong. The people who are still against
ATM on religous grounds should get over it.]

If service provider Y sucks, then customers will leave it.

Neither happened to UUNET.

2) if it was a success, the determining factor was its use of ATM,
   rather than its first mover advantage,
financial/management stucture,
   industry trends, etc. For example, UUnet used ATM because that was
   what the bellheads would sell them. At the time, there was no
   alternative to the bellheads.

First of all, UUNET got a lot of things right. As Craig said, they
were an excellent ISP. Those factors you mentioned were all important.
However a significant roadblock in any one area can slow down your
growth. At one time, UUNET's traffic was simply outrunning the
ability of L3 routers to keep up. Their choices were to:

  1. Adapt to another network architecture that scaled
  2. Have a poor network that did not scale, and therefore
      would drop lots of customer traffic
  3. Refuse customers that wanted to sign on.

Taking positions 2 or 3 would have caused them to lose market
share. They took position 1 and, combined with doing many other
things right, maintained their growth.

And as I already mentioned in another message, UUNET's network ran
directly over TDM. I don't know what you mean about what the Bellheads
would sell them.

Prabhu

At one time, UUNET's traffic was simply outrunning the
ability of L3 routers to keep up. Their choices were to:

That does invite a companion observation: at one time, I could reasonably
guess that their traffic was outrunning the ability of Layer 2 ATM switches
to keep up, as ATM switch vendors have lagged well behind the curve in OC48
and OC192 trunk capability, and IP router vendors haven't even put OC48 ATM
linecards on the market.

-travis

If service provider Y sucks, then customers will leave it.

Ahhh.. The lie of pure capitolism...

This is the assumption that the customer has perfect knowledge. It's
wrong. It's not as simple as: If they suck, the customer will leave.

growth. At one time, UUNET's traffic was simply outrunning the
ability of L3 routers to keep up. Their choices were to:

As you admitted before, this is not a factor in the market today. I don't
see much frame or ATM gear doing OC-192 switching.