The large ISPs and Peering

Thu Jul 26 03:41:41 UTC 2001

On Wed, Jul 25, 2001 at 11:03:48PM -0400, Vijay Gill wrote:
> Majdi, I don't believe I'm going to have to do this.

	I was in a hurry when I wrote this, and I probably should
have clarified a bit more:

> Perhaps this is why neutral (n-th) party colo providers were given the
> chance to bid on the contract. If the service is going to suck, it will
> suck for _everyone_. And there are plenty of people who are dropping
> equipment into competitors colos to provide service.

	I was mostly responding to the Level(3) in the poster's comment;
if something like this were to happen, Equinix and PAIX are definitely
the preferred vendors in this situation.

	I think the distinction should be made between colo in the sense
the original poster was using it (AboveNet, Level(3), etc.) -- colocation
provided by someone who would typically sell you transit along with your
colocation.  And colo in the PAIX/Equinix sense, where the vendor most
explicitly will *not* sell you transit, because they strive to maintain
a carrier-neutral environment.  Many small providers are not typically in
the latter (although some very definitely are), but certainly exist in the
former.  It's two different markets, so expecting that you are 'screwed' if
your carrier(s) of choice are not there is not really correct -- the carriers
will go where the customers are, regardless of where they might choose to 
meet.

	I realize I was not clear on this earlier :)

> Yes, but the long term traffic growth rate is constrained at the edges for
> the most part (edge being the local loop). Horror stories regarding speed
> of provisioning between two large promising local ISP's would fill a book.

	Yes, but dark fiber is a perfectly legitimate metro option.  The
cities mentioned in this thread are by and large major markets where dark
fiber is available.  You can do this without any joint facility.  5 years
ago, it may have made sense.  With the increasing availability of fiber in
these major markets, I'm not sure that the expenditures are justifiable in
the long term (beyond the next two to three years).

> So looks like some people have figured out that provisioning a dark fiber
> cross connect is much cheaper and easier to upgrade than say, getting new
> STM-4 and 16's from various people in n points across their topology.

	I'm not disagreeing with this, for obvious reasons.

> Maybe. You can play financial capex games on equipment etc.  Once you have
> this in place, it's a sunk cost; the upgrade path is clear and much easier
> than getting the ilec du jour (maybe another department) provision new
> circuits.

	Again, not disparaging the incredible usefulness of dark fiber.

	--msa