At the risk of sounding Chicken Little-like, we hypothesize that the Internet is heading toward a "tragedy of the commons". The tragedy of the commons plays out when a collective system is destroyed due to overuse. An example is the Boston Common that was set aside for communal livestock grazing during early colonial times. It wasn't long before overuse made it unusable for grazing. In the case of the Internet, consumer access networks are the commons—and they are oversubscribed. It's a fact of life that as we all use the Internet more, the oversubscription problem will progressively worsen.
Although we can't predict when overuse of the Internet commons will translate into a deteriorating experience for most broadband users—it will happen. And we believe that the Internet community should act sooner rather than later to define and implement mechanisms to protect user's interests. Notice we say the "Internet community" should do this. That is because an Internet tragedy of the commons is a communal problem that should be resolved by the community as a whole, not just a single stakeholder (e.g. an ISP).
Some people tell us: "It's all the ISP's fault because they oversubscribe their access networks and they should fix it by giving us all more bandwidth." While others tell us: "It's the fault of rogue users hogging bandwidth, and ISPs can fix it by controlling their behavior." We counter both of these viewpoints by repeating what we said a few weeks back—we are all in this together and all stakeholders must cooperate to hammer out solutions Last week we said that before we can solve Internet performance problems we need facts to identify and characterize each problem. With that in mind we ask—do you know of good solid data about whether or not we are heading toward an Internet tragedy of the commons? If so, please share it with us! We believe this information is important, and as with global warming, we expect the problem will be difficult to document at first. But if we don't understand the problem and take action to solve it, we may all be sunk.
NetForecast is an internationally recognized engineering consulting company that benchmarks, analyzes, and improves the performance of networked data, voice, and video applications.
The opinions expressed in this Weblog are those of the writer and may not represent the opinions of Network World.
|
|
I'm not familiar with the
I'm not familiar with the Boston Common but I'm assuming they didn't have the resources to just buy more land in order to feed more livestock. In this situation the ISPs have the demand AND the resources to avoid over-subscription. If one ISP chooses not to, another will. The users will jump ship and the first ISP will be forced to catch up in order to stay afloat.
While I don't mind likening internet users to livestock, there is a big difference between the internet and a big field. One is a business, the other.... is a big field. Bandwidth can be upgraded at any time. Grass needs to grow.
If I'm off-base on my interpretation of the Boston Common, let me know.
Commons is a Shared Essential Resource
The livestock in the case of the Internet is not the user but rather the software the user employs. The point is that most users have no idea how many network resources their software consumes. Much of the traffic that a desktop generates is not under the user's direct control. If the software is designed to consume as much network resources as it can, then it will.
Do you know how much traffic your machines are generating?
There are simply too many clever computers able to take advantage of the Internet principle that connections operate on an end-to-end basis. The computers (and their software) are in charge of consumption. Under these circumstances no carrier can ever build enough capacity to satisfy all the connected subscriber computers all the time.
Local access services are by nature oversubscribed. Every car can't get onto the highway at the same time, every phone can't be on POTS lines at once, and all commuters can't fit in the subway at the same time. So what makes you think that all the computers connected to an ISP can generate unlimited amounts of traffic at once without adverse consequences? The practical difference between the people-driven examples and the computer-centric example is that it's rare that everyone wants to get onto a highway together (unless there's a hurricane bearing down) so dramatic congestion events are the exception. Computers on the other hand can be connected to the network all the time generating traffic without human involvement.
The idea that if an ISP is oversubscribed then customers can switch to that other that isn't doesn't wash. In most of the U.S. there is only one credible broadband ISP, or at most two. If you jump from your cable operator to the phone company's DSL service you are just trading one limited pool of bandwidth for another. It's literally a pipe dream that a third carrier will appear that will supply an unlimited pool of bandwidth for $29.95 and compete with the cable and phone companies.
Like it or not you and everyone else are currently using a limited pool of bandwidth that you are sharing with many others, and you have no idea how much you or the other subscribers are consuming. That is why we believe that the local access network is a commons--and just like livestock, the software on your computer does not care about the health of the commons.
software using bandwidth
so the questions are, how DO you determine how much bandwidth you are using on an hourly and/or daily (or some time frame) basis when actively on the net and how much is being used by the computer when you are not actively on the net?
And then how do you compare and roll this number up?
And how do you minimize bandwidth use without impacting user experience?
and....
Software using bandwidth
I think that users can tell what bandwidth they are using. If you just turn on a PC and do nothing, the only apps that will be using bandwidth are going to be windows and other apps that update themselves.On PCs, you can use the task manager to find out how much bandwidth is being used. It may not be totally accurate but there is enough feedback.
If you pay to get on the internet, the natural thing is to want to get your money's worth. To use the internet commons idea, I bought a ticket that allows me to let my cows graze there, so I will make sure that my cows get as much grass as they can. ISPs would love to be able to say "Well we would like your cows to eat this much so we can allow more people to graze their cows".
ISP and users are not going to be having the same interests, ISPs want to have as much farmers with tickets and the farmers want to have as much grass for their cows. So you can see the ISPs wanting more limits and the farmers wanting less.
What is fair use of bandwidth?
This is made more complicated because ISPs want to have lower numbers and users want higher ones.
One trend is that some Tv Networks like NBC put episodes of popular shows online. NBC might not like to find out that people are not watching their content or paying for it because users have to look over their shoulders to make sure they are within their bandwidth limits.
Can anyone come up with a good number?
Remember that even websites are not as slim now like they were back then before Flash and other tools came out. Hit a few tricked out sites when close to the limit and you are in trouble.
Are users going to be happy to pay for internet service and then paying for bandwidth caps?
Now, how does one know if they are abusing their bandwidth?
Other than running a internet server if the contract forbids it or filesharing copywritten content, how else can a user abuse their bandwidth limits?
Right now we were probably just talking about Personal computers (PCs,Macs) but what about other devices?
I wonder if we give these stats to the ISPs are we giving them a stick to beat us with? Lets say that the ISPs find out that normal users 50% of the bandwidth.
As an ISP how much bandwith do you give users?
Lets say ISPs were giving 100 units and users were only using 50. For the most profit maybe you can give users 51 units or maybe 50. The problem is what happens when it costs more units to surf the net? The trend is for larger sites and for more things to do on the internet. So a number that was good for 2008 may not be useful for 2010 and later.
Will ISPs be smart enough to adjust for that or will they find it better to charge folks for overuse and get more money?
I have left out the other stuff like viruses and DRM and that needs to be put in the mix too.
Is the Internet like the Boston Commons?
Matt got it exactly right. The Internet is already grossly oversubscribed for what it was and it has consistently expanded its capacity to meet the demand as long as there was financial incentive to do so. There are also places in the market for numerous pricing models. In my neighborhood there are about six ISPs. One is the incumbent telephone company. It offers the lowest price, but oversubscribes its connection from its DSLAM to the Internet to the point that users may receive service only three or four times faster than a good dial-up connection during peak-load times. Others provide fast DSL at a higher price. Another has attempted to limit the service provided the user (sometimes in ways that have resulted in fines from the FCC.) New capacity including fiber to the curb is continually being added, and the cost of Internet bandwidth is continuing to fall making unlimited bandwidth cost models even more viable. Competition continues to force price downward. Technology continues to improve. The Boston Commons can indeed expand to be much larger than it is.
Post new comment