The Myth of the Bandwidth Hog Part 2

The Internet continues to grow very fast causing network operators to complain that flat rate; all-you-can-eat charging models are no longer appropriate. Many commentators assume it is inevitable that volume of data will have to be used as a basis for charging consumers extra. The theory is that broadband is the same as other utilities. Just as utilities all base their charges on consumption of the units delivered (water, electricity and gas) then so should broadband.

But there is a big problem with this comparison.

Those other utilities have two costs. They have the cost of building out the distribution network that delivers the gas, water or electricity. In this regard they are like a network operator.

They also, however, have the cost of producing the water, gas or electricity – completely unlike a network operator. It’s not that there isn’t a cost of production of network bytes. It’s that the cost of the production of those bytes is borne by someone else – the content originator. Google, Microsoft, Yahoo, Netflix and everyone else that produces the content have massive server farms that “create” the bytes.

There is also a more direct correlation between the number of bytes consumed and the cost to the content originators. As more bytes need to be produced more servers need to be purchased. The servers take up more space and need more power to run.

So, traditional utilities bear both the fixed cost of delivery and the variable cost of production for the units that are delivered. A network operator and a content originator share those costs. The network operator bears the somewhat fixed cost of delivery. The content originator bears the more variable cost of production.

There is another big difference between a traditional utility and a network operator – the rate of change in their costs. The delivery portion of a traditional utility actually increases over time because the equipment (pipes, cables etc.) costs increase as raw materials get more expensive. Manpower costs associated with construction are also increasing. The same is true for the cost of producing the units. The long-term trend for production of gas, electricity and water (mostly driven by increasing environmental regulations) is increasing.

But for both the network operator and the content originator the costs are falling – and falling fast. Moore’s, Kryder’s and Butter’s laws apply to all of the equipment cost elements. They all show a massive and continuing unit cost improvement. While this is partially offset by the increasing unit cost of manpower the overall cost is still trending down.

So where does all this leave us?

In part 1 I suggested that network operator’s costs are not driven by data volume. The number of customers that are trying to use the service during the busy hour drives their costs.

In part 2 I have suggested that the total cost of delivering[1] consumers the data they request is shared between the network operator and the content originator. And both the network operator’s and the content originator’s unit costs are falling very fast, and very predictably.

I believe that this sharing of cost is appropriate and should continue as both parties have revenue streams arising from what they do. It is also logical that content originators should have more of a unit based revenue stream (pay per view, per click, per ad). And that network operators should have more of a fixed fee per-consumer business model (pay per access pipe, differentiated by speed).

But, while the network use is growing extremely fast, as measured by bandwidth consumption during the busy hour(s), maybe the appropriate additional charge for a network operator to levy on its customer is use during the busy hour. If a higher percentage of their customer’s use occurs during the busy hour(s) then they should expect to pay more than people who only use the network when there is lots of unused bandwidth. Or if you want high speed during the busy hour(s) you pay a premium. If you don’t need high speed during the busy hour(s) you pay less.

[1] Note; I am only discussing the cost of delivering bytes here. I am not considering the cost of producing any content.

The following two tabs change content below.

Mark Taylor

I work as VP of Content and Media here at Level 3. English expat and passionate new tech energy evangelist.

Latest posts by Mark Taylor (see all)

7 thoughts on “The Myth of the Bandwidth Hog Part 2

  1. The costs of providing last mile access are not flat and certainly not decreasing. First, physical infrastructure requires maintenance and often rebuilding and this is true if you’re looking at HFC, twisted pair (telco), or FTTx. For example, cable plants had to be completely rebuilt to handle data upstream traffic cleanly in the 90’s and continue to have to get more and more efficient moving from 550 mHz, to 750 mHz, to 950 mHz to finally over a 1 gHz today in total transmission capacity. That takes both manpower and hardware not to mention lots and lots of physical cable. Much of the cost of providing access has nothing to do with electronics like CMTSs or DSLAMs or FTTx concentrators, which as you correctly note are following a decreasing cost per unit of processing. The cost of fittings, physical cable, “dumb” electronics like amplifiers, batteries, and the entirety of the physical plant are driven by the same commodity markets that you note are driving costs for the traditional utilities and in most cases the cost per foot is much higher for communications providers than it is for most utilities because the media is largely copper and has a higher manufacturing cost.

    Having said all of that, I do agree that the main drive to controlling costs needs to focus on peak usage, but that’s a very difficult concept to explain to end users(much less bill for) than a total bandwidth cap. The total usage cap is imperfect, but is more palatable to users than most of the variable controls I’ve seen in the market. Making sure they are accurate, clear, and reliable is something than any provider will have to tackle if they want to implement a system like that.

    • I agree it may be difficult but aren’t people comfortable with off peak pricing for electricity? And that was introduced for the same reason. Peak drives their cost and so they sought to rebalance those peaks a little. And yes I realize they charge for volume but the peak charging introduces a temporal aspect to that.

      • The short answer is no, they aren’t in my opinion. There are places where its mandated but the state PUC (I think Connecticut does this or used to) but where its not a regulatory requirement the RESIDENTIAL uptake is very very very low. Residential consumers hate unpredictability and since no one, including the service provider, can say when a peak time occurs or how long it will last its extremely unpredictable. In my experience even more savvy business consumers aren’t big on peak pricing for electricity and only very large consumers of power really spend the time and energy to analyze whether it would benefit them or not.

        • Scott I get that. I’d also say its more prevalent in other countries. But my purpose in these posts was to acknowledge that much of the consumer broadband industry (wired and wireless) is already charging for usage. And then to suggest that the method being used, although easy to understand from the consumer perspective, isn’t necessarily aligned with changing behaviours in a way that limits the need for network growth. Peak hour charging on roads and for electricity does try and do that. How successful that is is what you are challenging and I understand that.

  2. As a long time customer of level3 and even longer network operator, changing pricing on a per volume basis or putting a premium on peak hours is absolutely absurd since it hurts our business and our customers alike and not to mention our ability to attract more business to utilize our gig circuits at our datacenter or sell dedicated ethernet services direct to our customers.

    The carriers have the capacity, hardware, and budgets to handle high volume traffic almost effortlessly… this sounds like more the typical corporate greed to me than anything else.

    • Basic economics and market efficiency show that structuring pricing in a manner that accurately parallels costs yields optimum value for all involved. Customers that drive network expenses the most by presenting disporportionate peak loads during busy hour get to pony up more for the costly capacity increases that they’re driving. Sure, a peak usage cost component is definitely annoying, feels punitive and may even wreck a customer’s business case. It will however, drive responsible behaviors on behalf of all customers- such as rebalancing load to off-peak hours or compressing their peak hour content.

      Customers that aren’t driving up peak load see a considerable reduction in their rates. The carrier has very good reason to court them for their nearly zero marginal cost of service. These customers benefit since they’re no longer subsidizing capacity buildouts that they aren’t benefiting from. Overall, network utilization peaks flatten and the network operator’s costs decline. A cynic would assume the “greedy corporation” would pocket the savings. A cynic would be right absent greed-driven competition.

      In the face of competition, customers and a network operator together benefit from a cost advantage that the competition doesn’t have. Level 3 had such a product in the early days with their soft switch-driven managed modem dialup service. Customers and Level 3 benefited greatly from this architecture.

      If Level 3 adopts a peak busy hour pricing component, makes it understandable and prices it right with a low fixed component, then they should attract far more revenue than they will lose. The competition may get a few high peak customers that migrate looking for subsidization from off-peak customers, but there will be far fewer off-peak customers to stick it to on the competitor’s network.

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload the CAPTCHA.