Second-hand smoke from the Net-Neutrality pipe
I may regret this, but one thing I can’t abide is second-hand smoke. I haven’t seen one sane thought about the Comcast BitTorrent “fiasco” yet. The “people’s” argument is as follows:
Internet access should be unhindered.
I want to buy my Internet service from Comcast for $xx.yy per month; where x is min(x).
I want to use BitTorrent as a file transfer mechanism.
If Comcast interferes with my BitTorrent downloads they violate axiom 1, and are therefore pure evil and must be destroyed.
This is ludicrous. Granted, the way Comcast and Sandvine reportedly went about limiting BitTorrent, is not a good one: shoot down connections with TCP RST’s – gotta love the Great Firewall of China (er… Philadelphia). But there is absolutely nothing wrong with my provider making sure that everyone gets a fair, equitable, and secure connection to the Internet. In fact that’s what I want my carrier to do.
What Comcast is trying to accomplish (along with every other carrier on the planet), is how do they provide a good level of service to every subscriber at a pricepoint that has razor-thin margin. It’s not easy. Appetite for high-bandwidth applications are only increasing, they’re roughly doubling every year with no end in sight. However, the distribution of people that use bandwidth is not uniform. I have seen very real empirical data from an MSO near you, where something on the order of 90% of their traffic is generated by less than 10% of their customers. It gets worse, for that same broadband provider, almost half of their bandwidth is consumed by less than 0.5% of the customer base. Yes, you read that right. Guess which kinds of protocols are running 24/7 from that base?
Peer-to-peer file sharing protocols are not the best way to use network resources. They certainly may get the file to the client fastest; but they have evolved in a way that makes the original Internet architects groan. Many of these protocols utilize multiple concurrent TCP sessions pulling the same data from multiple servers, racing to see which one can get you the data fastest. Hint: your p2p client’s speed is coming at the expense of your neighbor. Or conversely, your neighbor’s p2p speed is extracted from your web traffic. But that’s OK right! Give everyone a network packet gun and call the network operators bums if they can’t keep up at $39.99 a month access costs.
I’m sorry, but Net Neutrality isn’t about letting my sixteen year old neighbor trash my UBR ring by up/downloading 100Gig of porn or stolen movies a month using p2p filesharing protocols. No! It’s about making sure that your Grandmother can get online to read her email, surf the web, and not get her machine owned in the process at a pricepoint that is within her budget. I’m sorry if that offends anyone, but that’s what we’re really talking about here.
Net Neutrality, isn’t supposed to be the wild west, where the person with the best client-side networking stack wins the game. It’s about equality, where the bully next door — who has hacked his kernel to ignore TCP congestion control — doesn’t drive your grandmother off the road
Have you seen a Cisco/Juniper 10Gig pricesheet lately? Do you know what the going rate is for a 10Gig linecard? Any ideas on how many lambdas are left on the transpacific fibers? It turns out that it is extraordinarily expensive to double a network’s backbone and transit speeds (let alone their access rates). Unless you want your Internet access pricepoints to grow at a rate that approximates health care costs, you will want to embrace technologies that your carrier can use that will actually guarantee per subscriber access on an equal basis. That’s what I call real Net Neutrality.