[Exchange-construction] Re: Ethernet-framed broadband

[Simon Blake]

Hiya Bill, Morten

On Fri, Mar 14, 2003 at 07:15:25PM -0800, Bill Woodcock said:
> 
> I'm cc-ing this to Simon Blake, who runs the most successful system that I
> know of similar to the one you envision.  If he has time, he may be able
> to give you more specific advice than I can.  Also, he's exactly on the
> opposite side of the world from you, so there's novelty in being told how
> to do something by somebody who's upside down.  :-)  I'm in Kabul right
> now, so I'm just sideways.

I could be anal and point out that exactly antipodes of Norway would be
somwhere like the Auckland Islands, or possibly even Antarctica, but
that would just be nit picking.  Madrid is Wellington's opposite, I
believe :-).

> Then you'll need GigE switches to form the core of your network.  Simon is
> using the Cisco 3512G switches, which I've now put in quite a few of as
> well, with good success.

I've still yet to find anything cheaper per GigE port, so yes, we still
use the 3512G's.  The reason for running GigE has little to do with
performance requirements - if you're using LRE you'd easily cover your
needs with 100Mb, but rather that the fibre termination gear (GBIC's,
versus external media convertors) is actually cheaper for GigE.  I think
I worked out recently that it's about US$750 for one end of a
100Mb fibre link, and about US$1000 for one end of a GigE copper link.
So make sure you go GigE, you get a 1000% performance increase for a 30%
increase in costs.

>     > future, more and more fiber will be ready for use. I will also aim to
>     > have my local employee, and a local company to administrate the
>     > project.
> 
> Be careful that you don't over-staff.  Money that goes into fiber and
> splicing labor and so forth is an investment.  Money that goes into
> administration is waste.  It's necessary, but should be minimized to the
> degree possible.

Damn straight.  Administrators have a nasty habit of turning into make
work experts.
 
>     > You mentioned an Ethernet solution for this kind of project. I have
>     > planned for use of the Ethernet standard. What do you think about the
>     > radio-link. In my opinion, this is a waste of money when we already
>     > have accsess to the copper cables.
> 
> Yes.  The thing wireless is good for is supporting mobile users like
> laptops and PDAs.  Also for temporary point-to-point links while you're
> still getting real cable infrastructure turned up.  But you run out of
> spectrum if you do that too much, and you have to deal with interference,
> and the equipment is relatively expensive, and it's not very fast compared
> to wireline.  So it winds up becoming a big time-sink, without much
> long-term payoff compared to time spent running cable.

FWIW, we (Citylink) have stopped doing ptp wireless completely. It looks
simple, but there's a deceptive amount of complexity and knowledge
required to get it going, especially in situations where there's
competing users (of the frequencies).  So focus on running glass.
 
> existing taps.  In Nunavut, they use solar cells up on poles.  I've seen
> places tha tuse the small wind-generators that are often found on
> sailboats, with a batter at the base of the pole.

sailboats can generate their own wind?  How cool is that :-).
 
>     > I also plan to implement ASP in the project. Do you have any good
>     > advice regarding this?
> 
> Yeah, don't do it.  That would be splitting your attention too many ways.
> Instead, get some young guys who wouldn't otherwise have high-speed
> Internet access to do that business and offer it to the customers.  That
> way you don't wind up having to cross-subsidize it from the infrastructure
> project.  You've got more than enough on your plate already.

Again, listen to Bill.  Run away!  Run away!

If you want to add value to the network, concentrate on getting as many
ISP's as you can onto it, offering service to the customers.  That way,
you don't have be an ISP, so you don't have to worry about usage
billing, so your cost models are lower.  Don't get involced with any ASP
stuff that could be perceived as a competitive product by people you'd
like offering their services on your network anyway.

Likewise, don't fixate to heavily on security, or QOS.  Just get lots of
bandwidth going, as absolutely as cheap as you can.

Here in Wellington, "we're providing an electronic footpath for the
city".  Anybody can use it for whatever they like, we don't charge them
for their use of it (only a flat fee to connect at a given rate).  All
we do is get people from somewhere in the city, to somewhere else in the
city.  If you want to get to the Internet (or wherever), you contact an
ISP on our network, and use them to get off our network.  We don't
restrict who you can talk to (it's a big ugly flat VLAN), nor are we
suggesting that it's secure or safe.  It is, quite literally, a big
electronic sidewalk.
 
>     > My biggest concern for now is how to present the project in such a
>     > way that the community will fully understand the benifits by local
>     > network that is partly owned by the county itself. If you have any
>     > good advice here, or any hints about how to make a good presentation,
>     > I will be delighted to hear about it!

Stay away from the techincal details, blind them with economic benefits
"if you do this, the county will save n million krona/year (or
whatever)"

Cheers
Si