First published on my blog Occam's Razor.
Upset by proposals by the Federal Communications Commission to create “express lanes” on the Internet? If the current proposal now out for public comment becomes a rule, it would allow Internet Service Providers (ISPs) like Verizon and Comcast to charge a fee to those web sites that want faster content delivery.
This is the opposite of net neutrality, which is the principle that all web content should be delivered by an ISP at the same speed. (Actually, it’s at the same bandwidth, since all network traffic is effectively at the speed of light.) The argument goes that without net neutrality, those companies with deeper pockets, particularly those who are already established, such as Netflix, have an unfair competitive advantage over other services or start ups without such deep pockets. It’s a concern I certainly share, so much so that I first blogged about it in 2006. Bottom line: I am still concerned and I think this proposal must be fought.
What I didn’t write about back in 2006 was that there was no net neutrality back then either. Effectively, bandwidth is already discriminatory because it is based on ability to pay. It’s just based on your ability to pay, not the content provider’s. For example, Verizon has basically four tiers of Internet service from it’s “high speed” service (actually it’s lowest speed service) where content delivery does not exceed 1MB per second to its “high speed Internet enhanced” service where you can download at up to 15MB per second. It’s hard to quantify what the cost of the 1MB/sec plan is compared to the 15MB/sec plan, because it depends on many factors including what bundle you may or may not choose. Suffice to say if you want a 15MB/sec service, you will pay more than a 1MB/sec service. So if streaming Netflix is critical to you, consider their 15MB/sec service. (Of course, this assumes that the port between Verizon and Netflix can handle 15MB. If it can’t then there is no point in paying Verizon the premium.)
You can think of the Internet connection from your ISP like a water pipe. If the water pipe is big (and the water pressure is high enough) you can get more water per second through a bigger pipe. What the FCC is proposing is to take this pipe and put two pipes inside it. One is a fat pipe that will serve certain content very quickly, the “fast lane”. The other smaller pipe is for those who can’t afford to pay ISPs these premiums, i.e. the “slow lane”. Since I live in traffic-congested Washington D.C., I think of the “fast lane” as the pricey HOT (High Occupancy Toll) lanes on the beltway, and the “slow lane” as the toll free and usually congested other lanes. It’s not hard to imagine the Internet feeling a lot like it did in 1995, when the hourglass is principally what you saw in your web browser. Pages took forever to load, if they ever did. For those of us who remember those days, revisiting them sounds quite frightful. ISPs would have every incentive to throttle the slow lanes, because it would mean that web content providers would come to them and negotiate to use their fast lanes. In addition, they would have little incentive to increase bandwidth for their customers overall, but plenty of profit to funnel back to stockholders from those that pay for fast lanes. It is the antithesis of what the Internet is about.
So already there is no net neutrality of content delivery, unless you have an ISP that provides a “one speed for all customers” plan. The issue is not content delivery; it is the speed of particular content distribution within the ISP’s network. Which brings up another less noticed way that the Internet is not equal. It has to do with Content Delivery Networks (CDNs).
If you access my blog with a browser you will notice it takes a while to render a web page. Why is that? It’s because I don’t pay for a content delivery network. I did a test from home on accessing my web site. I had to go through 13 routers (switches on the internet) between my home computer and my web host:
1 wireless_broadband_router (192.168.1.1) 0.525 ms 0.244 ms 0.216 msElectrons still travel at the speed of light, but they are thirteen stoplights between my computer and my web server, at least for me. You can see how long my request took at each stop. For example, hop 13 took 49.597 milliseconds. Add up all the milliseconds to see how long it took for me to get to my site. If you do the same thing, the number of hops will probably vary, along with the access time. In short, it’s relatively slow to get to, which alone may explain why my traffic is down. People are impatient when they click on a link to my site from a search index. So they go elsewhere or get an effective CDN by using a subscription service to read content like Feedburner or feedly.com.
2 l100.washdc-vfttp-93.verizon-gni.net (188.8.131.52) 7.083 ms 7.095 ms 8.161 ms
3 g1-5-3-0.washdc-lcr-21.verizon-gni.net (184.108.40.206) 9.435 ms 12.101 ms 12.305 ms
4 ae7-0.res-bb-rtr1.verizon-gni.net (220.127.116.11) 9.731 ms
so-12-1-0-0.res-bb-rtr1.verizon-gni.net (18.104.22.168) 27.151 ms
ae7-0.res-bb-rtr1.verizon-gni.net (22.214.171.124) 8.855 ms
5 0.ae1.xl1.iad8.alter.net (126.96.36.199) 10.166 ms
0.ae5.xl1.iad8.alter.net (188.8.131.52) 9.396 ms 10.254 ms
6 0.xe-8-3-1.gw12.iad8.iad8.alter.net (184.108.40.206) 9.610 ms
0.xe-8-0-0.gw12.iad8.alter.net (220.127.116.11) 9.693 ms
0.xe-10-1-0.gw12.iad8.alter.net (18.104.22.168) 10.872 ms
7 customer.alter.net.customer.alter.net (22.214.171.124) 8.733 ms 10.023 ms 9.717 ms
8 he-2-4-0-0-cr01.ashburn.va.ibone.comcast.net (126.96.36.199) 10.252 ms
he-2-6-0-0-cr01.ashburn.va.ibone.comcast.net (188.8.131.52) 14.819 ms
he-2-5-0-0-cr01.ashburn.va.ibone.comcast.net (184.108.40.206) 12.388 ms
9 he-4-3-0-0-cr01.56marietta.ga.ibone.comcast.net (220.127.116.11) 39.468 ms 42.618 ms 37.101 ms
10 he-1-12-0-0-cr01.dallas.tx.ibone.comcast.net (18.104.22.168) 42.852 ms 45.176 ms 44.283 ms
11 be-22-pe01.houston.tx.ibone.comcast.net (22.214.171.124) 50.270 ms 49.438 ms 50.270 ms
12 as8075-1.2001sixthave.wa.ibone.comcast.net (126.96.36.199) 49.692 ms 85.009 ms 50.379 ms
13 188.8.131.52 (184.108.40.206) 49.597 ms
This is not much of a problem if I go to google.com. Here is the route:
1 wireless_broadband_router (192.168.1.1) 0.557 ms 0.229 ms 0.202 msSo basically Google has figured out a way for its servers to be “close” to me, usually geographically, so I get their content more quickly, or at least with fewer stoplights between their servers and my computer. This magic is done through a content delivery network. I’m pretty sure Google rolled their own, and that takes a lot of money, which Google helpfully has.
2 l100.washdc-vfttp-93.verizon-gni.net (220.127.116.11) 6.919 ms 8.588 ms 7.432 ms
3 g1-5-3-0.washdc-lcr-21.verizon-gni.net (18.104.22.168) 12.248 ms 12.530 ms 9.252 ms
You can imagine if a company wanted to create a new amazing search index, it would be at a significant disadvantage if it didn’t have a content delivery network. They probably won’t roll their own like Google, but use one of the networks out there that do this for profit, like Akamai and Level 3. The technology behind this is interesting but I won’t detail it here. The linked Wikipedia article explores it if you are interested. Suffice to say it does not come free, but there are times when it is justified. The U.S. Geological Survey where I work uses a commercial content delivery network. Whenever there is a major earthquake they push the content out to the CDN, otherwise their servers would get overloaded and it would be like a massive denial of service attack. It also gets this data out more quickly to the public, as the typical customer probably only has to traverse three hops instead of thirteen to get the information.
We like to think that the Internet is free, but of course it isn’t. We all pay for access to it. Even if we don’t pay it directly, we pay indirectly, perhaps for the cup of coffee at Starbucks while we surf on their wireless network, or through taxes if we use Internet kiosks at our local library. Doing away with net neutrality is just another means by which ISPs hope to make gobs of money from having a monopoly on the last mile between the content you want and your computer. This may be due, in part, by our refusal to pay for their pricier tiers of service. The only difference is that this time you are not directly paying for it but other content providers will be. (You would think ISPs might cut you in on the deal and discount your rate, but that assumes they are benevolent, and not the profit-obsessed weasels they actually are.) As we all know, nothing is free, so these costs will certainly be passed on to you if you are a subscriber, and that profit will go to the ISP.
Given that bandwidth to the home is a limited commodity, giving discriminatory access to web content providers that can afford to pay must by necessity mean that others will get less access. In that sense, the latest FCC proposal is smoke and mirrors, and it is in everyone’s interest to get off our lazy asses and oppose it.