A discussion on one of my wireless ISP mailling lists caught my attention and I thought it was worth sharing.
—
On Mon, Sep 12, 2011 at 11:28 PM, Ken Hohhof wrote:
So the government has to foster development of killer apps that need >100
Mbps Internet?
http://www.nsf.gov/cise/usignite/
—The response from my good friend John Scrivner…
I guess that’s one way to force government backed fiber down everyone’s throat. I can see it now. They’ll start telling Grandma and Grandpa in a few years that if they want their social security check they have to apply from an application running at 1 Gbps or else they need not apply. That would be one “killer” app for sure!
John Scrivner
—
I have a problem with committing government and educational resources to discover new and innovative ways to consume more bandwidth when applications of this nature are going to be limited to a subset of our population and businesses. Let those people who have access figure out their own ways of taking advantage of their connectivity. Projects with this kind of a focus are putting the cart way in front of the horse. I sit in a place where a 100meg connection costs me $3500/month and a gigabit will cost $9000-$10000/month. The focus should be on dropping that cost and opening up fiber networks so that their true value can be unlocked and distributed to everyone in a way that is not dependent on government subsidy or regulation.
I feel that the twin pushes for “fiber everywhere” and cloud computing are dangerous to our society as a whole because of the culture of dependency that they foster. In a vacuum, fiber everywhere is not a bad idea, but in the real world it doesn’t make fiscal or practical sense. While a company like Google can do a one-off project like Kansas City, they cannot or do not take on the bigger challenge of putting that kind of connectivity into every village center in the US, and opening up access to that kind of connectivity to innovative providers who can deliver to the last mile. Instead, we have an entire culture of policymakers pushing for more government subsidy for telecom and broadband deployment that focuses on baubles like 4G (which is basically a toy), ultramodern but closed R&E networks that only benefit academia and maintenance of the money flow to the same telcos that have been holding us back. The Google Kansas City project is like a carrot held in front of a draft horse to keep the public distracted from pushing for policy changes that would improve the universal availability of better broadband at affordable prices.
Cloud computing is providing much of the money and motivation behind the efforts of so many individuals and groups that are pushing the idea of ultrabroadband. I don’t have a problem with cloud computing on its own, but becoming dependent on any outside resource of that nature runs counter to the ideal of self-reliance and resilient communities. I am an old school ISP, so I run my own servers and will not outsource critical pieces of my service to the cloud where it can be subject to the whims of an outside entity. That is my choice as a businessman and others are free to make those decisions on their own. I am fairly disgusted by the wholesale movement of our educational institutions away from the operations and maintenance of their own IT resources to cloud providers. Those institutions may be saving money now, but they are forgoing the educational opportunities for students and faculty to gain valuable operational experience on their own systems. They are also sacrificing local employment opportunities to enrich cloud providers. Cloud computing also weakens the resiliency of our computing infrastructure by serving as a huge target for disruptive conduct. Witness the regular breaches of corporate data and personal financial information. Bad actors within a cloud computing provider have access to highly sensitive information about businesses and people. Privacy policies and encryption are one thing, but in the end there are people in these positions of access and power – and people are corruptible.
We are duplicating the “too big to fail” philosophy that is proving to be a massive fail in our financial system by overemphasizing the need for ultrabroadband and cloud computing. I am much more interested in maintaining resilient systems that can stand on their own and improving the applications that are successful on our current broadband infrastructure while working toward the goal of universal broadband access even if that access falls short of ultrabroadband.
The first 1meg of broadband is far more important than the last 99.
Got a few questions for you.
Question 1: Do you host other people’s websites? If so, you could market that as a “cloud.” You don’t have to be Amazon.
Question 2: What colleges provide the opportunity for students to be sysadmins over systems that matter? Here at Colorado School of Mines non-student staffers take care of critical systems…which have gone down for hours at a time where cloud-based systems haven’t missed a beat. Specifically, our Blackboard course management site was down for ~14 hours straight a week ago, and prior to the transition from locally based e-mail (with a crappy Horde webmail UI and 500MB of storage vs. the gigabytes that commercial e-mail providers offer) we would have regular outages (try 99.7% uptime). My perspective: farm services that don’t need to be produced locally out to places that can do it better (keyword: better).
Question 3: How much spectrum do you use to power your access network? Verizon has 22MHz in 700, plus AWS holdings, PCS holdings and 850MHz holdings. Right now they’re using three of the four spectrum slices, and on 700 are hitting 30 Mbpds downloads and 10 Mbps uploads in real life. They have caps now, precluding people from using the service as a high-transfer residential service, but they’ll be starting fixed service (more spectrally efficient; you know that) soon which will have more liberal caps, better foliage penetration (that’s what happens when you buy really expensive spectrum) and higher capacity per MHz (no interference) than unlicensed. Not sure it’ll be such a toy at that point. It’s disconcerting for unlicensed wireless providers anywhere Verizon (or AT&T) might launch this service, but it’s best not to be in denial :/
Hello Ian,
Responding to your questions:
1) Yes, I do offer email and some other hosting services. They are in a NOC that I own and operate and have complete control over. I’m not dependent on some outside entity to do those things for me. Key word is independence. Cloud computing leads to dependency. Excessive dependency on outside resources breeds weakness.
2) Sounds like CSM could use better sysadmins. There is a learning opportunity on campuses for comp-sci students, where they can learn to deal with complex problems, management of sophisticated systems and even (horror!) dealing with end users. Those are potentially valuable experiences that are tossed aside when services are outsourced. Do our schools want to breed strength and resiliency or weakness and dependency? That is the question I am posing.
3) I’m using what everyone else is using – 2.4, 5.8 and some 900. I don’t really worry that much about what the other guys are doing as long as I can keep my customers happy. You should take a few minutes to read my white paper about fixed wireless deployments to get a better understanding of how the mobile networks are limited by physics. Putting a fixed antenna on an LTE system is not going to solve the problem either – any system that has to deal with mobile clients is going to suffer from loss of capacity when compared to a true fixed wireless system. I’m not in denial, I’m a realist. I’ve got 25meg service at my house on an unlicensed system and the new platforms are capable of delivering more than that depending on the system design. Add to that the fact that I can put up a basestation for $5000 that is comparable in performance to a $500,000 LTE base station and I’m even less concerned about what the cellcos do.
I should read your blog more often; thanks for taking the time to make your response to my comment. I’ll definitely take a look at the white-paper.
As for giving students the opportunity to work on systems, helpdesk (think Geek Squad) positions are readily available but that’s as far up the chain as things get :/ All the (Cisco) network gear isn’t touched by any student AFAIK