CNET logo Why You Can Trust CNET

Our expert, award-winning staff selects the products we cover and rigorously researches and tests our top picks. If you buy through our links, we may get a commission. Reviews ethics statement

Is bandwidth the new water?

The problem with not metering limited resources is that it can drive the wrong economic behavior.

Gordon Haff
Gordon Haff
Gordon Haff is Red Hat's cloud evangelist although the opinions expressed here are strictly his own. He's focused on enterprise IT, especially cloud computing. However, Gordon writes about a wide range of topics whether they relate to the way too many hours he spends traveling or his longtime interest in photography.
4 min read

In the 1980s, New York City undertook a huge project to begin metering residential water service and charging for individual apartment and condo water use rather than simply assessing a flat fee as had been the norm. It was as huge and contentious a project as one would imagine. Why go to all this trouble and expense?

Well, you hardly have to be a radical free market economic thinker to accept that there's generally a connection between how much something costs and how much people consume. The higher the price, the less you use. In this case, the incentives have more to do with installing water-efficient fixtures, fixing leaky faucets, and the like. But the idea is the same. In short, New York wanted residents to start conserving water and realized that financial incentives were the only practical way to make this happen.

On the enterprise IT scene, I'm seeing the same sorts of issues replayed around datacenter power consumption. Most everyone professes great concern for how "green" (to use the trendy lingo of the moment) their operations are. In practice, if the power bill isn't in their budget (and is therefore effectively a flat rate of $0) nothing much usually happens until someone higher up who is responsible for all the relevant money flows makes it so.

But incentives for efficient datacenters is a topic for another day. Let's talk broadband instead. There's considerable ruckus brewing over whether it's right and proper for ISPs to meter their broadband pipes to at least some degree. It's a contentious topic--and one on which I find myself much torn.

I have no particular love for the telcos. Customer service tends to range from mediocre to awful, a problem only exacerbated by the common case where one provider is "the only game in town." (I was reminded of this just a couple of weeks ago when trying to track down a phone line problem.) And they have a history of tacking on all manner of add-on charges--leading one to suspect that usage caps or metered usage are just the latest such.

Furthermore, I, like just about everyone else, prefers the predictability of flat-rate pricing. Especially when it's easy to imagine that network bandwidth consumption could well be much more variable than electric or water bills. It's not randomness that led dial-up ISPs to switch to unlimited plans essentially en masse. I remember counting my (expensive) on-line minutes at the old Compuserve; there was nothing customer-friendly about that.

And what of all the new video services such as Amazon's Video On Demand (nee Unbox)? Do we really want to see the pricing structure of the pipes get in the way of consuming such services?

At least some accept that perhaps--just perhaps--there's some justification in price-discriminating against true outliers--those who consume hundreds of GBs of bandwidth per month. Whether it's really worth all the noise and heat doing so generates is another matter. (See fellow CNET Blog Network's Peter Glaskowsky's discussion here.) But, many others then argue, what of a future where a mainstream user is today's outlier? When video is routinely delivered 24x7 in high definition across the network.

Alas, that's where the rub is. I would sincerely like to see Moore's Law and other technical advances cheaply enable the sort of bandwidth increases that would make this possible. And my best guess is that they will--eventually. I was attending conferences well over a decade ago where online delivery of just-about-everything was "right around the corner."

At the same time, it may also turn out that--for at least some years to come--that the network really isn't the most cost-effective way to deliver certain types of content. As Andrew Tanenbaum's book Computer Networks so quotably reminds us: "Never underestimate the bandwidth of a station wagon full of tapes hurtling down the highway." Netflix has effectively applied this dictum to the delivery of video.

Over time, it will make sense to deliver the bits that represent more and more types of content over the wire. And "the industry" works every day to make that a reality. But we all also need to be practical about distinguishing what makes sense today versus what makes sense in the future. A big part of that is for both consumers and providers to react to the economics of things as they are, rather than as we wish they were. This will ultimately be the most effective way of encouraging the right kind of investments so that we can move further and further into the digital age.