NNSquad - Network Neutrality Squad

NNSquad Home Page

NNSquad Mailing List Information


[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[ NNSquad ] Re: Why Gigabit DSL matters

This is how "Moore's Law" is supposed to work and why I've argued that we're
underutilizing our copper (and wireless and fiber) assets.

What is important is that we can use any technique whether it is finding the
one pair that can do 1Gbps or using smarter electronics/software to glom
together to get the capacity it doesn't matter. The point is that there are
many approaches and that if "the market" sees an opportunity it will find a
way. Note that the whole "speed" thing is confusing -- the bits go out the
same speed but we can exchange more at a time if we take advantage of
opportunities. This is why I prefer to use the term "capacity".

We should stop talking about FTTH or any presumption about technology and
use what is available and by normalizing to bits we can mix and match in any
combination that works.

The FTTH (et al) requirements have worked to maximize costs while making it
seem difficult to build out connectivity. This is a double whammy in that
the high costs make it hard to justify providing capacity whereas
incrementally using available physical infrastructure can be far more cost
effective and drive demand. Furthermore by focusing on speed within billable
paths we minimize the capacity we get after all the effort.

The "physics" of information can be weird. A single gigabyte path can do far
more than provide 10 users with 100 megabits each because of statistical
sharing. Each one can get more like a gigabit each. Even if you use all 10
are using 10Mbps streaming there is still lots of capacity available. The
more capacity we aggregate and share the more effective capacity we get.
Divvying up the capacity into pipes creates "breakage" or the loss of
capacity by making it unavailable to others when unused.

The Telco business requires channeling bits into billable channels for
billing purposes even if it means, in effect, throwing away the vast
majority of the capacity due to breakage. An unshared 1Gbps path may average
out to 1Mbps (or far less) which means that the cost of billing requires
throwing away more 99% of the capacity of the unshared path. At very least
consumers should join together to share the path themselves.

We have abundant physical capacity -- it's the provisioning that is problem
and that's policy not physics. Just as the bandwidth shortage is a construct
due to policy and not due to a limit on our ability to exchange bits.

More at the usual essays: http://rmf.vc/ThinkingOutsideThePipe,
http://rmf.vc/sd etc.

-----Original Message-----
From: nnsquad-bounces+nnsquad=bobf.frankston.com@nnsquad.org
[mailto:nnsquad-bounces+nnsquad=bobf.frankston.com@nnsquad.org] On Behalf Of
Lauren Weinstein
Sent: Sunday, December 18, 2011 13:20
To: nnsquad@nnsquad.org
Subject: [ NNSquad ] Why Gigabit DSL matters

Several NNSquad readers reacted to a posting of mine yesterday
(regarding a Gigabit short range DSL tech), suggesting that existing
Ethernet standards provide the same functionality.

I believe this DSL technology is of most interest where utilities are
trying to leverage existing copper cable plant, especially in
underground utility situations (which often can mean 50+ year old
cables in hard to access locations buried directly in the ground).

I would expect any successful high speed DSL system to generally have
better noise immunity and crosstalk rejection characteristics than
conventional Ethernet -- especially important in old, tightly packed
cables that can have 500+ pairs.

But the big advantage in these situations is only needing one pair,
vs. two or four for 100bT or 1000bT Ethernet.

Fewer pairs means fewer interfaces and nominally less expensive
equipment, but the big win relates to the fact that many of these hard
to replace cables are so old and so badly maintained (and/or in poorly
documented bridging configurations) that many/most of the pairs won't
even test out suitably for data.  Only needing to find one good pair
for a customer is a win in these kinds of situations.

So I think there is a quite valid place for this tech.

NNSquad Moderator
nnsquad mailing list

nnsquad mailing list