last mile capacity [was Re: QOS or more bandwidth]

hardie at hardie at
Tue May 29 17:24:46 UTC 2001

This approach has the interesting property that it requires the agency
engaged in the "nationalization" to have at least approximately the
same reach as the area to be served.  Unless it is truly the national
government, there are lots of cases where this does not work in the
U.S.  Think of the D.C. area, where you have multiple U.S. states,
counties, and municipalities interrupted by a capitol territory
controlled directly by the Congress.  This is probably a worst case,
but even in the San Francisco Bay Area (one state, no funny national
territories) there are 7 counties and dozens of municipalities.  They
do not currently share basic municipal services (fire, water, sewers)
and the few coordinated services are not noted successes.  This is one
of the reasons that things like the Sewer Access Module fiber builds
are slow in the Bay Area--the sewers don't interconnect from, say,
Palo Alto to Menlo Park (which are contiguous and indistinguishable to
the casual observer).

The Stockholm and Montreal fiber builds are good examples of what can
work when the agency involved does have appropriate reach, but it may
not be the correct way of solving the problem when you have the urban
blur that is common in the U.S.
				Ted Hardie

Sean Doran writes:
> The correct way of solving this was demonstrated in Stockholm
> and duplicated in a handful of Canadian cities.   In the first case,
> the City of Stockholm "nationalized" the laying down of dark fibre
> in the city, and formed an agency ( which 
> provides unlit/unrepeated/unamplified dark fibre between any pair
> of addresses in Stockholm at cost as a public utility.
> Thus, instead of a dozen or so CLEC-style companies ripping
> up the same set of streets, Stokab does it approximately once,
> and provides fibre pairs as necessary to these companies,
> and any other buyers who come along (lots of corporate buyers
> use Stokab instead of the traditional telcos or CLECs).
> This approach has been an unqualified success for Stockholm,
> which thanks in large part to Stokab's establishment in 1994,
> has been *the* intersting place to do Internet stuff through most of
> the years since then, despite the city's geographical remoteness
> and small population.
> The major drawback of existing dark fibre utility agencies is
> their management's tendency to try to be innovative - Stokab for 
> example sometimes appears (misguidedly!) to want to move up the 
> value chain into services their buyers are offering, and into new 
> experimental things involving media other than fibre (e.g. radio).  
> If a single "nationalized" supplier of dark fibre slows down or
> becomes more expensive as a result of this, it will cease
> to be a market-enabling success, and start to look like the
> sort of constraint on the last-mile market that former PTTs
> are imposing on their captive market.  (And then yeah you're
> back to digging up more streets or using QoS or whatever, sigh.)
> So, a good idea is to press your local government into duplicating
> Stokab (it really IS good for you), but stop your local equivalent
> from ever hiring someone with a bell-shaped-head or technology fetish.
> 	Sean.

More information about the NANOG mailing list