Re: RE: [REPForum] REPforum get-together

> > The MAE's were not provider-centric, they were exclusively 
> > for providers. They were also not required by the R&E 
> > communities, nor did the NSFnet "break up."
> 
> OK, my semantics are a bit off, but the general point remains. The MAEs were recommended as a way to keep the tail sites interconnected while using competitive backbone providers. People were afraid of balkanization.

I still think you're understating it.  The creation of the NAPs and
the transition to them in 1995 was a fundamental change in architecture
of the Internet itself.  ...anyway, it's not that important to the
conversation.  :-)

> > As far as these transit exchanges are concerned, larger 
> > providers simply
> > refer to these as LEC's.   The typical customer wouldn't understand
> > this exchange point connection type of deal, unless, perhaps, 
> > you are talking about a univerity connecting to Internet II
> 
> We're starting to see more mid-to-large orgs interested in these exchanges. Emergence of community dark fibre networks has a way of bringing them out. I think you might have to be closer to these (potential) clients to begin to see an emerging trend.
> 
> EG, Teraspan here in Vancouver can get you a fibre ring (say 3-5km, pair) in the downtown core for 1500-2000 CDN/mo. But try to get that thing terminated in a CO. No way. But you can get it terminated at the transit exchange easily and cheaply. So I
> guess the LEC is Teraspan in this instance, but there's only GBICs on the ends, and those are customer-controlled. Really pretty simple.
> 
> BTW, fibre in Prince George in the downtown core is even cheaper, especially when downtown is about 8 square blocks or so. There are at least 4 similar municipalities with solid plans for fibering their downtown cores for local businesses. All of them
> are looking at some sort of transit exchange for service provision to end customers.

Okay, I see a metro fiber company looking to have a carrier hotel as one
of the points on their ring so that all those potential customers they
went by in the street have somewhere on their network to go to.  They're
a LEC, or provide services to LEC's.

> So now they're terminated with GE at the exchange switch, and the user wants to multihome. And peer locally. Hopefully the exchange operator can facilitate this.
> 
> BTW, this can also be done with L1 at the patch panel and multiple fibre pairs. No switch nec. But this lacks scaling efficiencies.

What is inefficient about buying one port per customer?  Someone has to 
pay for it...so you make the customer buy from a "neutral" provider's 
switch, and you have to convince the customer that you're the best 
choice for transit.

How about you just walk into the customer's office, tell them you can
sell them a GigE port with a 10meg commit at an unbelievable price, 
local loop included, and call it a day?

I think your target market is much bigger if you look at SME customers
that can afford a switch with GigE then if you look at larger customers
that are ready to do GigE into a router and manage BGP sessions.  The
market just can't be that big.

> You are right, the increased performance is a benefit. But users are starting to see value in connecting to an exchange as an end-point for dedicated condo fibre builds, obtaining service from multiple providers, ease of changing providers, and peering
> with other similarly connected metro organizations. Like it or not, this is more user/consumer driven.
> 
> Is there a sustainable business model here for exchange operators, providers, and users? Open question.

Your situation may vary.  I have product managers coming to me every couple
of months asking me to build into locations like these.  Most of them
can't write up a positive business case to justify going in there.  As
a large carrier, the way these things typically work is we justify going
into a location, e.g. Equinix, based on reducing local loop costs on
peering.  Once we're there, we can sell services to customers as they
show up, but rarely do we show up based on customer committements alone.

I suspect it's the same for a smaller ISP.  There's got to be a serious
level of pressure to spend your hard earned dollars from customers on
a presence at an exchange if it the cost savings and/or revenue 
opportunities are uncertain or intangible.

Take the scenario you pitched above.  If you are on a metro fiber ring,
the exchange is on a metro fiber ring, and the customer is on the 
metro fiber ring, why do you need to meet the customer at the exchange?
Why not just sell to customers on the ring?  If you're not on the ring,
get on the ring to get access.  Maybe that's where the exchange comes
in...a good spot to colocate where they're racks, space, battery backup,
etc.  Even so, you don't have to admit to the customer that you're there,
just sell your stuff!

So, I still don't see anything that says "be at a regional exchange
to sell more transit."  That being said, if your customers are savvy
enough to know what a regional exchange is, and want to be at one
and buy service from you there over their existing link, I guess you'd
better look at the business for that and make your decision.  :-)

Dave

-- 
Dave Siegel
HOME   520-877-2593   dave at siegelie dot com
WORK   520-877-2628   dsiegel at gblx dot net
                      Vice President, IP Engineering, Global Crossing