Comments on LARIAT and Comcast not same problem

[David Farber <dave () farber net>]

________________________________________
From: Brett Glass [brett () lariat net]
Sent: Saturday, February 16, 2008 9:41 AM
To: David Farber; ip
Subject: Re: [IP] LARIAT and Comcast not same problem

Dave:

For IP, if you will; some responses to a few comments which folks have made on the list regarding LARIAT and network 
management.

Dave Burstein writes:

> Brett's comments are perfectly appropriate for a small U.S. rural wireless carrier, and totally off base for a large 
> DSL or cable provider like Comcast or Verizon.

Actually, they're equally true for both. In fact, in many ways the big providers are hit much harder than I by P2P. 
They have larger upstream pipes at the head end, and larger downstream pipes through which to receive requests for 
content. This multiplies the impact of P2P, which relentlessly attempts to consume every shred of available bandwidth. 
When we send a recalcitrant P2Per to the cable company for service (and, yes, we do this if they're within the cable 
company's territory), we joke that it will hurt them much more than it hurts us to lose the customer.

Several folks on the list have mistakenly asserted that wireless networks face constraints that wired ones don't. This 
is only true, however, if the networks are badly designed. We design our networks to have FEWER constraints than the 
cable plant, which has issues due to protocols such as DOCSIS. Expansion is easier for us, because we don't need to 
string new cable or fiber (though I do need to spend an afternoon dangling from a tower or building). And our network 
is symmetrical; DOCSIS isn't.

> Protecting rural service is a battle worth fighting (possibly directly excluding small carriers from some regs, 
> unbundling rural fiber backhaul, or efficiently using the $300M in the current) but it shouldn't be used to make bad 
> policy for everyone.

It certainly would be useful to get some regulatory relief and/or some compensation for our hard work. But allowing 
providers to fight network abuse is good policy for all providers, large and small.

> Since so many of the FCC filings got the facts wrong I'd like to point out a few things I believe proven. These do not 
> force a pro or con on the subject.
>
> It is possible to run a large broadband network without traffic shaping like Comcast, because Verizon, AT&T, and 
> Free.fr do it.

Actually, according to the comments filed with the FCC by AT&T and Verizon, they both do it also. See their comments, 
which are excellent reading (and required reading, IMHO, for anyone commenting on the subject:

Verizon
http://fjallfoss.fcc.gov/prod/ecfs/retrieve.cgi?native_or_pdf=pdf&id_document=6519841190

AT&T
http://fjallfoss.fcc.gov/prod/ecfs/retrieve.cgi?native_or_pdf=pdf&id_document=6519841106

> Internet growth rates per subscriber are little changed for the last five years at 35-45%, not increasing or a crisis 
> brought on by video. Source: Odlyzko's excellent MINTS page and Comcast's filing with the 40% number for 2007
>
> Over that same period, the costs of delivering that bandwidth have gone down at a Moore's Law pace of 25-40% 
> (switches, routers, etc.)

Unfortunately, the cost of backbone bandwidth has not gone down -- due, primarily, to extreme consolidation in the 
industry (the SBC/AT&T merger, Level3's acquisition of Wiltel and Broadwing, etc.). The quotes we're getting this year 
are higher than we got in 2005.

> My conclusion, which allows one to be in favor or opposed to NN, is that neutral networks aren't free, but are 
> practical at a definable price.

If you define "netural" as "zero traffic management," this could not come at a price that's acceptable to consumers. We 
estimate that our prices would have to start at $60 per month for just a moderately fast residential connection 
throttled only by total bandwidth (rather than being shaped). And users would still complain that their VoIP wasn't 
working when a P2P application was running.

It is amazing to me that so many folks who have zero actual experience operating an ISP or managing a busy network 
believe that they can pontificate about our costs and tell us how to run our business.

Andrew Burnette writes:

> Today?s average residential broadband user consumes about 2 gigbytes of
> data per month, Kafka estimated, which costs the service provider about
> $1.

Unfortunately, this estimate is provably extremely low. 2 gigabytes is about the volume of updates which Microsoft and 
Symantec, together, make to their products in a month. Thus, automatic updates ALONE cause any user with a single, 
off-the-shelf PC to consume that much per month.

As I mentioned in my filing with the FCC, our margins are razor-thin: about $5 per residential customer per month. One 
tech support call can lose us that entire margin, which encourages us to make our network ultra-reliable. And we seem 
to be succeeding, despite the handicap imposed upon us by the forced use of "wild West" Part 15 spectrum. We average 
one tech support call PER DAY, usually about a problem that is in the user's computer rather than in our network.

--Brett Glass, LARIAT.NET


-------------------------------------------
Archives: http://www.listbox.com/member/archive/247/=now
RSS Feed: http://www.listbox.com/member/archive/rss/247/
Powered by Listbox: http://www.listbox.com