connectivity outside the US

Sean M. Doran smd at clock.org
Sun Jun 1 01:14:24 UTC 1997


| Worldcom and Cable & Wireless are in the process of laying a 20 Gbps

Thanks for the ad.  Now to try to make this more operations-relevant...

Vadim Antonov and I had a brief exchange wherein I pointed out
that the biggest costs (noted in part by others) of trans-oceanic
cables are, principally, insurance, rental of cable-laying fleets,
ongoing management and repairs, acquisition of landing rights,
and restoration-bargaining.

SDH (I know of no trans-oceanic SONET deployments, however people
have been discovering that some ADMs don't care while others light
up like Christmas trees when one is used instead of the other) helps
reduce the costs of bargaining for restoration capacity, since a
properly engineered route _ought_ not to fail across both paths
for the most common reasons, however, even really well planned
SDH loops will fail from time to time.  This sort of technology
(TAT-12/TAT-13 for example) is a huge plus.

CWI has an "in" on fleet rental and on ongoing management, however
it does not reduce the price to zero as CWI and its partners
(e.g. SPRINT in PTAT Systems Inc.) can attest.

Landing rights to London are relatively inexpensive.  However, London
isn't the only place in Europe, and it will be interesting to see how
Worldcom manages keeping costs down on end-to-end international private
lines between the U.S. and Europe.

The technical aspects of the Gemini cable deployment
between NYC and London are staggering.  I expect that clever
ISPs (UUNET anyone?) could arrange trial services on the 
first path when (if?) it becomes available at the end of 1997
or the start of 1998.   However, it would be insane to bet
one's business on the cable being fully reliable for some time,
and doubly insane to bet on not needing a full restoration contract
in place until several months after the second path comes up 
probably about a year later.

AFAIK there are no practical means yet to acquire as much restoration
capacity as there is bandwidth on Gemini and until other paths
are brought on line an operator should consider acquiring backup
either at the IP level or by spending real $ on a restoration 
contract. 

What would be dead keen, however, is people getting together
to use whatever is available on Gemini initially as a size-large
battle net to test out the next generation of IP routers and the
like in a relatively public fashion. 

| All this means that we should see a significant drop in trans-
| atlantic bandwidth (and even intra-europe) charges in the near
| future.

Intra-Europe charges will fall for many reasons beyond MFS's
deployments.  I would be interested in a formal and binding assertion
that the Gemini owners will substantially undercut their competition's
pricing structures, otherwise, frankly, this seems like the kind
of hubris one shouldn't plan real networks around.

Finally, for the "bandwidth-is-free" crowd, this only amounts
to maybe 8 STM-16s (maybe more if people take chances and parallelize
across the two paths rather than use the two paths for redundancy),
which means this cable is really not that far ahead of the apparent
bandwidth-demand curve.  Consequently, don't expect the current
model of being shy of taxable short-term high-capacity deals to stop as
Gemini becomes available.

On the other hand, that CWI is willing to continue taking large
risks in developing private submarine cable systems and is able to
find investment and risk partners bodes well for being able to
continue terrestrial growth, and maybe alot less well for LEOs and
similar technologies.  This is good news for people building parts
of the global Internet.  Of course, a cream-skimming path that's
being built, so perhaps I overestimate the degree of risk perceived
by the two Gemini partners...

	Sean. 



More information about the NANOG mailing list