Beaten Up in the Blogs
Wednesday, December 05, 2007They say any publicity is good publicity. In this case, I say publicity is good except when it comes from those who have never studied physics or taken the time to run a few experiments with wireless connectivity. Most recently, I have been beaten up by comments posted on TechCrunch (scroll down to comments starting at 85), Blogger Joe Duck, who seems to think I have entered into senility early and, of course, Eric Schmidt, the CEO of Google who basically said I did not have a clue about bandwidth and its ability to grow with demand.
But then, writing things that some people don’t like comes with the territory―I use facts to expose flaws in conventional wisdom. So, I don’t mind being beaten up now and again, and it certainly doesn’t hurt my business. It only makes me realize that those who are convinced that our wireless resources are unlimited have never come face-to-face with the realities of bandwidth constraints.
And those who complain about pricing on a bandwidth allocated basis don’t understand that most of us are already making that same choice for our DSL or cable connections. For example, if I were using Cox Cable where I live, I would have the choice of 12 Mbps down and 1 Mbps up for $56.95 per month for cable internet access (if I also subscribe to cable TV, $64.95 if I do not) or I could have 5 Mbps down but only 512 Kbps up for $41.95 or $49.95 per month. If I was a light Internet user, I could opt for the value package of $26.95 per month and get 1.5 Mbps down and 256 Kbps up.
My Verizon DSL pricing follows the same model: 768 Kbps down and 128 Kbps up for $17.99 per month or 3 Mbps down and 768 Kbps up for $29.99 per month. If I wanted to use Clearwire’s WiMAX wireless broadband (if it was available here) I would pay $59.99 per month for 1.5 Mbps down and 256 Kbps up, $36.99 for 1.5 Mbps down and 256 Kbps up, and for $29.99 per month I would get 768 Kbps down and 256 Kbps up (so much for high-speed, high-capacity WiMAX).
This certainly looks like bandwidth related pricing to me, yet when I mention that a wireless network operator is contemplating this same type of pricing plan, or one that might be even lower but then has additional charges for downloading large files or streaming video, I get beat up. I guess it’s okay to pay for bandwidth in the wired world but not in the wireless world!
Try this experiment: Use your DSL line at home, connect one computer and do both an up and down speed test (repeat it several times and take the average) using http://www.speakeasy.net/speedtest/, which seems to be about the best and one of the most accurate speed measurement sites around. Now add a second computer and start downloading some streaming video and run the speed test on your other machine again. Okay, now add another computer that is also streaming video and run the test again. You will see a big difference in download speeds and in the amount of bandwidth you have for the first computer (keep these numbers handy).
Even Eric Schmidt dismissed my comments about limited bandwidth during an interview on NPR a couple of weeks back, maintaining that we will grow our wireless bandwidth to meet the demand. So, one more time, let me restate the issues as I see them:
- Shannon’s law and the laws of physics cannot be broken—bent a little perhaps, but not broken.
- Wireless bandwidth is a finite resource. We cannot make any more of it, we can only use it more efficiently, which has occurred with each evolution in technology so far. BUT there is finite limit based on Shannon’s law.
- If a network provider is using all of its licensed spectrum, the only way to increase capacity is to build more cell sites closer together and add pico, micro and femto cells. A standard cell site takes years to get through the approval process―not the weeks it takes to put in a new fiber drop.
- The cost of backhaul for wireless networks is running about 40% of their total operating expense and this will probably go up with the next generation of technology. There are a lot of cell sites that cannot be upgraded to fiber backhaul and for which microwave is not a practical solution either.
- Look at the specifications for the new-generation technologies LTE and UMB. You will see that they offer a significant improvement in data speeds in the same bandwidth, but remember that the more customers there are on a specific cell sector, the more users there are to share this total bandwidth. And someone who is downloading streaming video is using a lot more of it than someone who is checking email.
Now, let’s duplicate the tests you ran above with your DSL connection at a place that offers Wi-Fi access and has a single access point that is fed to the Internet using a T-1 connection (1.54 Mbps). No matter how fast the access point is (802.11b, g or n), the maximum speed you can get is the 1.54 Mbps minus overhead. If you are the only person using that access point, you have all of the available bandwidth if you need it. Run the same speed tests you ran on your home DSL service and you will see what is available to you when you are the only person using the access point.
Now add five or six more users, all doing normal web surfing, email, etc., and take your speed measurements again. Your results will show that your data speed has slowed, albeit not by 1/5th or 1/6th, but a lesser amount because while you are all sharing the connection, most of you are not downloading or uploading at the same time. Now, have two of these users start downloading streaming video and see what happens to your data speed. Until they stop downloading video, your data speed will be much slower because more of the bandwidth you are sharing is being used by those who are streaming data.
The same tests in a single cell sector of a wide-area network, even a WiMAX network, will provide the same results because you are competing for a specific amount of bandwidth in that cell sector and the more people who want to use it, the slower the data speeds will be. I don’t know why this concept is so difficult to grasp. Even when we are using wired or wireless voice phones, we are all competing for a finite amount of bandwidth.
The argument I get is that new technology, more cell sites, more backhaul and more spectrum will solve these problems, therefore, we don’t need to worry about or address them. This may be true to some extent, but there is still a finite limit to our capacity. Those who say that wireless will be deployed to meet the needs are missing two important points: 1) It takes a long time to develop, permit and build cell sites, even pico and micro sites, and 2) the addition of these sites cost someone money. Those who believe we will never have to worry about bandwidth also believe that demand should bring down the cost of service so it is almost free. Where will the money for increasing bandwidth to keep up with demand come from? Especially considering that these people believe wireless network operators should simply get out of the way and be dumb wireless pipes and not worry about any other aspects of the wireless ecosystem.
I know my rants here probably won’t change many minds, but I will feel better. I have tried, once again, to help those who want to understand with ways to experience bandwidth and speed constraints on their own with nothing more than a couple of computers. Run the tests, see the results, and then let’s talk about how much capacity we really have available. And we can talk about what happens when people want more than their share of it because it is their right to be able to access anything and download anything.
One final point, according to TeleGeography Research, Internet bandwidth grew more than 40% in 2004 through 2007, and in 2007 it will grow at a rate of 68%. Average Internet traffic will be up by an estimated 60% this year, so the Internet build-out is keeping pace with the increased traffic. If wireless data traffic grew 60% in a single year, network operators would not be able to match the rate at which wired and fiber providers increased the capacity of the Internet.
To those of you who are complaining about wireless operators using price to help manage the amount of bandwidth they have available, what’s the big deal? You are already making these choices based on pricing and speed for your wired connections. Why not wireless?