Paul Kapustka's Blog

You're not a network engineer, are you Paul?

If you have a really highspeed backbone, as Internet2 has, the access pipes are the bottleneck. When you upgrade them, the backbone becomes the bottleneck and you have to upgrade it, and the you're back where you started.

QoS manages a finite pool of bandwidth, regardless of the size of the pipes.

The AP story is substantially correct. Packet networks are all over-subcribed, that's their whole point.

Um, Internet2 backbone is at 10Gig. Think that will sustain for a bit.

My point is that the AT&T argument is all wet -- we don't need prioritization, just need upgrades to the pipes. Is it the users' fault that telcos are relying on oversubscribing as a business model?

And -- as I've posted before -- without transparency into the telcos' service offerings, there's no way of telling that QoS is even being used. That is why there should be regulations, just like for other commodities, so that customers can be assured they are getting what they paid for (and what the service providers are advertising).

We just need upgrades to which pipes? And who pays for them?

The Internet is a packet network, and all packet networks are over-subscribed by design. Over-subscription is not an accident, it's not a conspiracy, it's not wicked capitalists stifling freedom, it's the essential idea around the design of packet networks.

It's real easy to tell if your network supplies QoS or not: you examine packets with the appropriate tool, and you can measure performance. Network engineers to this stuff every day.

Upgrades to all pipes. If there is real competition for the right-of-way to my house, I am confident that a service provider will pay for the cable/fiber/antenna install in exchange for a committed (year, 2-year) contract from me. That's what happens now. Once on the Internet, I can go anywhere I want, pick the services I choose. I am also confident that content providers will want to spend on their infrastructure to ensure the best delivery, lowest latency, etc., possible.

Richard, you ARE a network engineer -- read your bio -- so why not entertain the idea advanced by Bob Kahn, among others, to build "public" COs where any service provider who wants can offer a service? Or Reed Hundt's idea to connect every business and residence to the net via fiber? Estimated cost, $20 billion. Cheap compared to what we spend in Iraq.

Hundt's plan isn't for the government to build it -- but instead contract out to whoever makes a competitive bid. Let AT&T build it, and keep 50 percent of the pipe, then open up the other half for real competition. Why not entertain some alternate technical ideas instead of just saying "net neutrality bad?" Or on saying that all networks have to be oversubscribed? Oversubscription seems to be the idea that has led to the current not-enough-bandwidth problem.

NetWare over Unix sounded like a good idea too, didn't it? Too bad that new idea wasn't embraced in time to see if it could work. And we all remember the success of the AT&T NetWare Connect Service! There's real QoS for you.

C'mon, Richard, be part of the solution. Think different, suggest ways to improve the network without trusting the telcos. Sure, we could all run network analyzers every day and sift packets on our own, like real men. Of course, we could all go to the supermarket and test the beef and chicken for mad cow. Or, we could all test the gasoline to make sure the Ethanol levels are correct. Sometimes, regulation isn't a bad thing.

Netware doesn't have much to do with any of this, one way or he other. It was a fine little system that performed well, but its day is done.

Regulation not informed by knowledge and experience is generally a bad thing, Paul, and the network neutrality movement is driven by fear and ignorance.

Packet networks are over-subscribed by design. They're successful because they're cheaper to build than the old circuit-switched networks, and they're cheaper because of the very resource sharing that leads to meltdown if they're over-used. It's a tragedy of the commons thing.

Any big network needs to manage bandwidth to keep resource hogs from ruining everybody elses day. That's life.

Hmm, no response to "what about new types of networks" question. Guess we need to find others to have an educated discussion on new networks.

As to Network Neutrality being driven by fear and ignorance... actually it's being led by some pretty smart and confident folks, including my employer, Jeff Pulver. Wouldn't call him fearful or ignorant. Jeffrey Citron at Vonage -- you may not agree with his business model, but ignorant? Nope. Not fearful either. But he's calling for a Broadband Bill of Rights because he's seen blocking of his company's service first hand. And the rule that let the FCC punish that service provider doesn't exist anymore.

Michael Powell -- originator of the Internet Freedoms ideas. Last we checked, he had a pretty good grasp on these things. Larry Lessig -- pretty smart there, wouldn't want to call him ignorant in public. But you go right ahead.

Vint Cerf? Also a smart person. And even Alyssa Milano -- who you, among others, thinks it's OK to make fun of -- a successful actor and small businessperson who is using the Internet to join the debate. But go ahead, make fun of her because she's popular. That's a valid way to argue, right? Funny, right?

As for packet networks -- since you ARE a network engineer, can you tell me how networks in Japan, Korea, Europe, etc. etc., are offering 100 Mbps symmetrical to their users? Are those oversubscribed? Or are they just big enough (for now!) so that provisioning and QoS don't matter so much? Are folk there being kicked off the network for "using too much bandwidth?"

And... am I a "resource hog" if I want to use the entire 7 Mbps downstream that Comcast sells me?

I would have no problem if Comcast, Verizon, etc. admit that their networks can't support their advertising, and change their messages to "buy access from us, and we can't tell you how much bandwidth you can have." I'm sure the prices they'd be able to charge would fall too.

Still waiting to hear what the better alternative is!

Paul, it's not going to be productive to discuss new networks with you until you get a clue about how today's networks operate, and why.

You're all surprised to learn that HDTV streaming has the potential to choke this network that we have today, and shocked that it doesn't have the bandwidth to instantly pump 1000 times more traffic.

Gosh.

You mention some smart people, some dumb people, and some people I've never heard of such as this Pulver character. It's not the case, however, that the people you list as supporters of your side are actually in favor of Internet regulation.

Michael Powell, for example, has come out in opposition to the Markey Amendment; Lessig has been silent on it, and Lessig's boy Tim Wu argues against that sort of thing in his Neutrality essay. Vonage is just asking for non-blocking, which COPE gives them, and Cerf is simply doing Google's bidding, as one would expect since Google pays him, so that pretty well just leaves you with Alyssa Milano.

I'll stipulate she's a hot little tart, but I don't see her name on any networking standards.

Richard, glad you have anointed yourself keeper of those who know all about networks. I guess I don't see Ed Whitacre's name on any networking standards either but since he has billions (and who knows how many "consultants" on his payrolls) I guess he can play in your sandbox.

In the meantime, since you still feel compelled to show your own simplistic take on things maybe I can help broaden your education:

On HDTV: When somebody is out there offering multiple HDTV streams, please let me know. Odds are the content will carry caveats, like most Internet video used to (some still does, asking users to pick download speeds to help ensure a quality experience). And odds are if someone really has a commercial reason to provide HDTV over a small pipe, they might try to do something on their end -- content caches, multiple geographical servers -- Bittorrent for non-streaming content -- to help matters. Point out to me where I said I was shocked by anything, other than the Bells' ability to obfuscate. I heard AT&T's side, and I heard the Internet2 side, which (and this was a network engineer talking) say that at 100 Mbps to the desktop they found they didn't need QoS or any special provisioning to do streaming HD conferencing. That, I guess, would be from a smart person by your definition.

Powell isn't against Markey or Net neutrality per se (last time I talked to him), he just thinks it doesn't make sense to battle the Bells head-on in a lobbying fight. Have to agree with him there. He also said he wasn't happy with the way the FCC de-fanged his freedoms by adding the caveats about how the rules matter unless the network operator decides they don't. Unless you've talked to Michael more recently than I have, you're wrong.

Lessig may be "silent" on the same amendment but it's hardly the only thing being said on the topic. In fact, he gave a very impassioned presentation at the most recent VON show in San Jose, the event run by Jeff Pulver and our company, pulvermedia. (If you need help using Google to find "Jeff Pulver," let me know and I can probably talk you through it even though I'm not a network engineer).

Unlike the Bells, who get all their troops to march lockstep, people on my "side" seem to be interested in finding the best way forward, not one way or the way AT&T says is right. Can't really answer for Tim Wu but he's just one of a bunch of smart people who have given this a lot of thought. Another is Martin Geddes, who actually argues against net neutrality legislation, but is very in favor of open networks. Maybe the idea of multiple people having different opinions is confusing to you; it's so much easier when everyone just picks a side, right?

And Vint Cerf -- well, gosh, all he and Bob Kahn did was pioneer all this packet networking and TCP/IP stuff you seem to know so much about. Funny, they never mention your name when I talk to them. But yeah, Vint probably just spouts whatever Google wants him to -- instead of the other way around, like Eric Schmidt saying he defers to Vint on this argument because Vint knows this stuff cold. Vint, by the way, was one of the people who thinks the AT&T/QoS video plan is "crap," something he told me last week when I interviewed him. So can he argue with you, if I'm not qualified?

As far as Alyssa Milano goes, I've not talked to her so I'm not sure how much she knows about networking. But thanks to your last comment about her, we can all see what a cretin you are. Back to the cave, dude.