December 22, 2009
[berkman] Brett Glass on the life of an ISP
Brett Glass is giving a Berkman lunchtime talk. Brett runs a Wireless Internet Service Provider (WISP) in Laramie, Wyoming. “Lessons from Laramie: Broadband Innovation on the Wireless Frontier” [his slides]
|
NOTE: Live-blogging. Getting things wrong. Missing points. Omitting key information. Introducing artificial choppiness. Over-emphasizing small matters. Paraphrasing badly. Not running a spellpchecker. Mangling other people’s ideas and words. You are warned, people. |
Brett begins with a photo of him on a 50′ tower on top of a 6-story building, which is what you have to do to be a WISP in Wyoming. He’s going to talk about what it’s like to be a WISP. “My entire life brought me to this,” although he didn’t know it as the time. He had worked at TI designing chips, and got a masters at Stanford. He was a computer journalist for many years: more than 2,000 published articles. He moved to Laramie because he likes it there. But the only way to get the Internet was at the University of Wyoming which had a couple of T1 lines, or Compuserv at 2400 baud. So, he founded LARIAT as a user group that turned into a non-profit cooperative ISP that bought its own T1 (for $6K/month). He got early circuitry to build wireless connections for the other users. But the members eventually just wanted to be customers, not coop members. So, Brett and his wife took LARIAT private in 2003. He has 18 yrs of experience deploying high speed wireless Internet. It’s growing by about the size of Manhattan island every year.
“Our cost to deploy is less than $100 per sq mile” at DSL speeds. But FCC regulations prevent him from beating cable speeds. The latencies are lower than DSL and much lower than satellite. The big choke point is that he can only get to the backbone through the telephone company, which charges them 10x as much as the telephone company charges to bring it to the rest of the world. They can’t get licensed spectrum because it’s too expensive, and unlicensed has too much noise in it; every baby monitor and wireless mic can disrupt their signal if they’re close enough to the access point.
Fiber is $25/foot. Brett can cover 40 sq. miles for less than $3000 in capital. “Nobody has this sort of cost per sq mile except perhaps the satellite guys.” “We turned out to have invented a really really good way to cover rural areas, and it’s pretty good in urban areas.” He’ll do things like go to the guy with the “trophy house” on top of the hill and offer free wifi so long as they can also transmit from that site.
The primary constraints on his ability to provide coverage and to innovate are political, not technological. “Fiber is nothing but wireless inside an expensive tube. The physics are the same. You just put it in a tube to exclude interference.”
He bid on 700mz auction, and even a tiny sliver to cover a tiny bit of WY went for $3M. There’s no way, he says, the telco that bought it is going to make a profit by using it; they bought it to exclude other players. “The incumbent has so much to lose that they’re willing to bid many times what the independent operator can.” Brett would like “lightly licensed spectrum”: Not polluted by consumer devices. I.e., a wireless broadband band. “It seems perfectly reasonable but national policy doesn’t have anything like that.” He says that white spaces coming available aren’t right for this.
Q: Doesn’t the 1996 Telecommunications Act provide for competition?
A: Short answer: We watched the courts remove the provisions.
Q: The newly-free analog TV bands?
A: This is beachfront property spectrum. Down the frequency it’s easier to penetrate walls. But the FCC did the wrong thing: They allowed consumer devices on the band, which means they can go through the walls and interfere with your neighbors. But there were powerful lobbies. People didn’t think about the science. Only the politics. It should have been reserved for transmitting from towers to get through walls.
“I had a customer I served on the same band you use for your wireless PC. This customer his service was failing on bright sunny days. We came over to his house when he was having trouble. Nothing was blocking his antenna. But on the patio his daughter was sunbathing in a bikini, with a cordless phone positioned right in front of the antenna. Both were on 2.4gH. We got her a phone on a different phone and that solved the problem.”
And, he says, we need to look at regulating only when there are real problems to be solved. E.g., getting inexpensive backhaul to the Internet. But all the focus in DC is on addressing problems that don’t exist, e.g., Net Neutrality. “No one is engaging in the practices they’re talking about outlawing. But no one is addressing the problems that actually exist.” [paraphrasing throughout!]
He shows a graph of Shannon’s law: To get a lot of bandwidth to someone, you need the frequencies to be uncluttered. Get too far down “Shannon’s knee” and the ability to get broadband falls off, but above the knee, returns diminish as noise goes down. This is an argument for cognitive radios [unassigned frequencies negotiated in real time by transmitters and receivers], which Brett’s been working on for years, but doesn’t have spectrum to implement.
Policy recommendations:
Devote more nonexclusively licensed spetrum to wireless broadband, with mandatory “etiquette” to enable cognitive radio. E.g.,. use the 700 MHz “D” Block. Take back the “virtually unused MDS “A” band for local wireless.
Q: Use it or lose it licensing, like liquor licenses?
A: You do that when you want to restrict something, keeping them to a zero-sum number.
Next thing to do: Increase power limits on when the devices are doing broadband in rural areas, not when they’re a baby monitor, etc.
Increase the geographic granularity of spectrum licenses so “little guys” can bid on them. Right now, the FCC offers narrow slices of frequency for large geographic areas; the FCC ought to do the opposite.
We should have a doctrine of “adverse possession” for spectrum. Right now, we have a feudal system when it comes to spectrum. New doctrine would be one like homesteading: If you use the spectrum, you have a claim on it. This is to get past the current policy of hoarding.
Do not write regulations that assume only content proviers are innovators or that only ISPs can be gatekeepers. We should be more worried about Google. “I have lots of competition in many of the areas I operate.” There are 3 WISPs in Laramie. “You don’t have to worry about me being a gatekeeper. But if you can’t get something listed on Google, you’re basically off the Internet.” “It’s easy to hate the big guys. But not all of us are big guys.”
Finally, “fiux the broken middle mile (special access) market.” There’s no competition there. “ISPs have never filtered any content” except maybe with a couple of exceptions, including filtering VoIP but that was taken off the table. Some of the regulations would outlaw some of the most popular service plans Brett’s company offers. He offers basic service for $30/month [rate plan]; for that you’re not allowed to operate a server, because a lot of the bandwidth Brett buys is asymmetric. People can buy a business class connection that does allow more bandwidth.
Q: Who are you buying access from?
A: The local exchange carrier, the only game in town. The only other way is to tap into the Level 3 backbone that runs along the highway. Level 3 doesn’t want to open up service to Laramie, except for hundreds of thousands of dollars + a guarantee of $15,000/month right from the start, which is way beyond that LARIAT makes. And microwave spectrum is too crowded to enable to get to the backbone in Cheyenne. The government should consider telling the carriers that unless they drop their wholesale prices below their retail prices, there could be some federal price caps.
Q: Is Level 3 quoting you that fee because they have sweetheart fees with the incumbents?
A: I can’t speculate. But others have told me that Level 3 only wants to sell to densely populated areas, even though we’ll pay them more per megabit.
Q: You can’t run p2p apps on your network, right?
A: Yes, because they’re servers.
Q: Title 1 Section 1 of Comms Act says that all radio frequencies ought to be used for emergency use. Section 303G [?] requires the FCC to make the best use of the spectrum. Have you considered a First Amendment suit against the FCC to challenge its right to regulate spectrum since there is no longer a scarcity?
A: I don’t have the money to sue the federal government. Interesting idea.
Q: What do you think of LRE, etc.? Or are you happy with 802.11?
A: 802.11 is not ideal for outdoor use. It’s designed for when all the transmitters can hear one another. But it’s good in that environment anyway. Wimax at it’s best is maybe 10% better, and sometimes worse. We tweak 802.11 to make it work better.
Q: Spread spectrum?
A: 802.11 is sometimes spread spectrum [Techie answer. Couldn't follow.]
Q: [me] Your $30/month plan provides a guaranteed minimal connection rate? What is it?
A: We started as a coop, so we had to provide transparency. We have always offered a committed information rate: You can receive and send a particular amount of info. The plus is that if someone is really a bandwidth hog, I can tell them that this is what they’re buying for that price. For $30, you only get 256kbps. Might double during off-peak. We guarantee this by buying enough. We monitor that someone can’t cut into someone else’s guaranteed amount. And we prevent people from doing bursts; everyone gets a VPN tunnel that controls how much bandwidth is allocated to each tunnel to make sure it gets its share and can’t hog. The New America Foundations nutrition label for bandwidth isn’t a bad idea, although it can be hard to figure out what you should be measuring.
Q: Muniwifi?
A: There’s no such things as a free lunch. Very expensive to run a high quality broadband network. You can’t just give it away. People expect to be able to stream and do latency-sensitive things. The amount of taxes you’d have to pay is about what you’d have to pay to a private provider. And they expect it to be ubiquitous, but 2.4gH won’t go through many walls.
Q: Would you consider doing this in developing countries?
A: There’s an advantage. They’re often not deploying telephone infrastructures. You could locate net service with the cellphone towers.
Q: LTE?
A: It’s just a buzzword for whatever we develop next. Long Term Evolution is the cellphone companies’ word for whatever they build next. Watch out for the hype. No one can violate Shannon’s law.
Q: Do WISPs need to be small? Could they get big enough to get over your scale problems?
A: There are some roll-ups. Some achieve economies of scale. But they don’t come to our area because they couldn’t get the backhaul. And you’d really need a good person in each city, and this knowledge is not easy to come by. And it’s hard to get investors; they walk away when there’s a hint that we might get regulated. I’d love to do it, though.
Q: In Cambridge we have a single ISP…
A: There are some WISPs in Cambridge. WispDirectory.com. There are 4,000 in the US. They hide because they’re worried about being squashed by the big guys.
Q: Why aren’t the other WISPs lining up behind you, supporting your ideas?
A: WISPs are very independent people who climb up on rooftops.
Q: Are there common areas where you can work with people on “the other side”?
A: Depends what you mean by other side. I’ve worked with CTIA. But WISPs are small, homespun businesses. “We’d like to see ourselves enabled, rather than hobbled.”
[Posted without re-reading. All possible errors and stupidities are mine, not Brett's.]
