Think Bandwidth Caps Are Unnecessary? Check Out This Graph
A Seattle man is furious that Comcast unceremoniously cut off his broadband — and banned him for a year — after he busted through Comcast’s 250-Gigabyte usage cap two months in a row.
Comcast’s two-strikes-and-you’re-out rule, without any chance to appeal, does seem draconian. (You can read the ex-Comcast subscriber’s rant on his blog here.)
One thing is clear: Broadband service providers will all need to do something to contain the rapidly rising flood of Internet data.
This is a historical graph showing the bandwidth utilization of Midcontinent Communications’ 100 Gbps network serving the Dakotas and Minnesota from January 2004 to June 2011:
It’s a safe bet that virtually every other Internet service provider could show you a similar-looking graph.
According to Midcontinent, Internet bandwidth consumption among its customers increased 116% in 2010 and has grown more than fourfold since 2009 (see Midcontinent Brings 100 Mbps To The Dakotas, Minnesota). Its network utilization is now at 25 Gbps of traffic; if that quadruples sometime in the next few years it will have to provision more upstream bandwidth and build faster backbone links.
Midcontinent would not provide specific subscriber numbers but said residential data customers grew 9.7% from the end of 2009 to the end of 2010. In 2011 year to date residential data subs are up 1.9%.
So far, Midcontinent has decided to refrain from imposing bandwidth-usage caps. But VP of technology Jon Pederson says “like most network providers we have evaluated this possibility.”
One of the conspiracy theories about bandwidth caps in general is that incumbent providers are trying to stifle Internet video services, like Netflix. In fact, Netflix’s general counsel, David Hyman, said just that in an op-ed recently in the Wall Street Journal.
“This anticompetitive aspect is particularly apparent when one stops to consider that AT&T’s U-verse is a television service delivered as Internet data traversing a network. Similarly, Comcast is testing its own Internet-based television platform in Massachusetts,” Hyman wrote. “So it’s no surprise that bandwidth caps would not apply to the data — e.g., TV shows — that AT&T and Comcast are delivering via broadband, but only to a third party’s data — e.g., TV shows from Netflix.”
Comcast has said in the past that video downloaded via its Xfinity.tv site does count against the usage cap. In Canada, Shaw says its streaming movie service also will go toward the monthly cap allowance. But the specific question Netflix’s top lawyer is asking here is whether managed IPTV services should be subject to the same maximum-usage limits applied to Internet bandwidth.
I don’t think so. Here’s why: The video going over the private, managed IP network is using dedicated bandwidth, allocated for that purpose. So it doesn’t cause congestion for other Internet services (like Netflix or Hulu) and vice versa.
Perhaps your feeling is: Hey, my ISP advertised unlimited, super-fast Internet access — so it’s up to them to invest in a network that can deliver that to me 24/7. It’s not my fault streaming video is eating up more of the pipe.
But there is a real cost for boosting capacity to meet nonstop video streaming. Netflix argues that the cost to deliver an incremental Gigabyte is “less than one cent, and falling.” In aggregate, however, it’s a different equation. Growing network capacity from 100 Gbps to 1 Terabit per second isn’t just a couple of pennies.
Hyman does suggest that limiting speeds by time of day would be an effective traffic-management approach. But why slow everyone down? Maybe enforcing usage limits based on time of day — and giving those who want to use a lot of bandwidth during primetime hours the option of paying more — could be a solution here.
What do you think? Add your thoughts below.
——————–
Follow me on Twitter: @xpangler
Ed Garman commented:
This graph, by itself, proves exactly nothing.
No surprise that peak usage is growing quickly. But peak usage appears to have grown from a small fraction of capacity to 25% over seven years, during which the service provider has collected monthly service fees from its subscribers with minimal marginal cost.
Most companies reinvest in their businesses without so much whining.
Chuckie commented:
Peter - seriously, that's your argument? If that was 10 years ago, that was before Netflix, Bittorrent, YouTube, Hulu.
Peter commented:
A 250 GB monthly cap in 2011? I aided a company in 2001 that had a 150 GB cap for ADSL. And they were very profitable.
Considering the decline in backbone capacity costs a 1.5 TB cap today would be very profitable and a more reasonable cap would be like 10 GB per month.
It
Chuckie commented:
Mark Aitken: What? Are we still talking about bandwidth management policies?
Mark A. Aitken commented:
Thanks Todd, I did see the followup from Cringely. And he is correct. We need new, innovative solutions. The problem is that right now, it is easier to take the lazy man's way out of a problem when you are a really big guy (40 pound monkey versus 800 pound gorilla). We just need to round up enough monkeys, and take on the gorilla together! ;-)
Todd Spangler commented:
Thanks, Mark. Did you see Cringely's follow-up, in which he responded to the owner of a small ISP who called his previous post "misleading and inaccurate"? http://www.cringely.com/2011/07/internet-class-warfare/
Mark A. Aitken commented:
Sorry, the web site I 'sighted' got dropped. Try this:
http://www.cringely.com/2011/07/bandwidth-caps-are-rate-hikes/
Mark A. Aitken commented:
Bandwidth caps are rate hikes
Go to this (), and you (too) will learn that that the talk of this being 'necessary' is (to quote an 87 year old mom) "BS".
Todd Spangler commented:
Phillip,
Consumers don’t like paying highway tolls, either, but transit authorities have found that those are a useful means for funding infrastructure and/or managing congestion.
To answer your question about subscriber growth: Midcontinent does not disclose specific subscriber numbers but the MSO said residential data customers grew 9.7% from the end of 2009 to the end of 2010. In 2011 year to date residential data subs are up 1.9%.
Regarding your point about Midcontinent’s marketing - that’s marketing. The reality of network utilization is another thing.
As you pointed out, Midcontinent *IS* investing in upgrading its network, as are all ISPs. My point is, if the usage trends continue on their current trajectory, something’s got to give. You either try to rapidly scale up demand to meet 100%+ year over year growth, or you use a mechanism like bandwidth limits to manage that growth.
Phillip Dampier commented:
This is more of Todd's wishful thinking in favor of usage caps. Consumers do not want them, and there is nothing about Midco's Scare Chart that sells them.
This chart is like trying to assemble a puzzle with 500 pieces and we've only been given five of them. We know nothing about what Midco's customer growth has been, their ARPU growth on premium speed tiers which naturally generate enhanced usage, whether they are maintaining an appropriate level of investment in their network to keep up with natural demand, and what their connectivity costs are. We'll never know, because Midco is a private company.
But here is what we do know. Mr. Pederson, on Midco's own website, indirectly obliterates most of Todd's points. He sells us on the fact Midco has enormous new capacity made possible with D3 upgrades, sells and promotes its own multimedia content add-on packages to its broadband service, and literally encourages customers to use the myriad of devices they connect to their Internet service. Indeed, Pederson seems to be aware the Internet is not used the same way it was in 2004, as some providers out there seem to be thinking.
When Midco sells 100Mbps service for over $100 a month, do they really expect customers to use that for e-mail and web browsing? It seems to me providers are keen to pocket that added revenue while balking at the increased usage higher speed tiers bring. And based on a lot of financial reports I am reading, providers are DECREASING investment in their networks, which means they are not even trying to keep up with demand in some instances. That's when we start hearing promotion of usage caps and speed throttles. They expect people to use a dozen gigabytes a month and that's it. When they use more, the Whine and Cheese Festival begins.
I've expanded on this considerably more here: http://bit.ly/pLm0HQ
Watch Mr. Pederson's own videos on this subject. Either he is lying to his customers, or Todd picked the wrong guy to use to promote his pro-usage capping positions.
In fact, outside of providers, the dependent trade press, and Wall Street, nobody wants these Internet Overcharging schemes.
Phillip Dampier
Editor, Stopthecap.com
Elgarak commented:
At the same time, the ISPs a often also delivering cable TV. Guess the amount of data of delivering a HD TV channel 24/7!
You can do the math, but it's about 6,600 GB at slightly less than 20 Mbps for a month. Multiply for each TV served with each account. That's what ISPs CAN and DO deliver routinely. Any cap that's lower than this for any customer who just pays and wants Internet data is screwing over the customer.
CanAmFam commented:
hey fung0, just wondering which incumbent cable/telco provider you work for...
fung0 commented:
It's important to note that current 'exponential' growth in demand IS going to level off. After video, there's nothing bulkier that consumers need to move. Furthermore, video largely represents not new demand, but a switch from older networks to new IP transmission, which should be more efficient, not less.
Chuckie commented:
@Tim - I understand what you’re saying and yeah - the ISPs *MUST* expand capacity. My point really is that there’s a big, big difference between 20%-30% annual growth in bandwidth usage and the more dramatic 100%+ levels seen lately.
Tim McElgunn commented:
Chuckie - OK. please compare bandwidth demand at dawn of cable modem era to today.
Now examine investment.
Now please graph cable co. income, free cash flow, and cash flow margin over the same period.
Ugly picture, ain't it?
Do you really believe you can price bandwidth sufficient to restrict demand without killing demand? Or, to put it another way, reduce your product's value to the consumer and keep them as customers?
If it becomes impossible to afford the bandwidth required to access and effectively use the applications/content consumers want, what do you expect them to do? (see Dialup, DSL, see ATTW)
Stop using the advanced apps? Say goodbye to innovation, innovating companies, jobs, and the ability to pay for anything near state of the art broadband.
I'm not trying to pick a fight, I honestly don't see what choice network ops have, except to invest and build. And, so far, that seems to have worked out OK as a strategy for increasing the value of the companies that do so.
Kevin commented:
It's clear that some kind of price adjustment will take place. Consumer demands do not always equal reality. Capitalism and true supply and demand will produce a winning formula for all...eventually
Todd Spangler commented:
Thanks for the discussion, everyone - some good debate here.
Wanted to call out New York Times editorial on this topic from Thursday — skeptical of the need for caps, and urging regulators to keep an eye on the industry:
To Cap Broadband, or Not - http://nyti.ms/o66p6B
Chuckie commented:
Tim - do you run an ISP? I don't think so, but hypothetically if you did, would you be prepared to spend to meet exponentially growing demand over the next 5-10 years? You would have a tough time selling that to your bosses.
Tim McElgunn commented:
*Unless you keep throwing more capacity at the problem (which is not a long term solution)*
Can you explain what else any telecom provider has ever done over the long term?
This is called running your business. Is cost per bit of capacity going up? Or is the same old curve continuing as the gear evolves (and cable op ability to drive prices down only improves)?
If the problem is in the backbone, all of the cable ops better stop selling commercial services right now if they can't afford to upgrade.
Bandwidth demand has exploded over the last decade. Anyone notice MSO free cash flow declning?
Chuckie commented:
as always, I blame Canada
fung0 commented:
There is no bandwidth crunch. This was made abundantly clear at Canada's CRTC hearings last week, in testimony from several large national service providers. Quotes from Bell and Telus, available in the official government transcript, make it clear that caps and usage fees are driven by competition, not congestion. Let's hope the debate can now move on.
Chuckie commented:
Phil - This shows **total** bandwidth consumed. The "so what" is that there is a finite amount of aggregate bandwidth available and if this exponential trend continues the ceiling will be hit and everybody's Internet will suck. Unless you keep throwing more capacity at the problem (which is not a long term solution)
Phil commented:
So, this is a graph of usage. So?
How about a graph of utilization, relative to the ever-increasing speed of the hard-wired infrastructure? I suspect you'd find a stepped, but mostly flat, graph.
Statistics...
Todd Spangler commented:
Mark - you've hit the nail on the head: that infrastructure for greater capacity must be built. There are costs associated with that.
Mark commented:
Once infrastructure is built, isn't cost/gb essentially fixed? At the moment there are huge profits in broadband. Rather than throttling usage with caps, broadband providers need to invest a portion of their profits in the bandwidth needed to meet consumer current and future demand. Today's heavy user will be the average user in a year or so.
Mark commented:
Once infrastructure is built, isn't cost/gb essentially fixed? At the moment there are huge profits in broadband. Rather than throttling usage with caps, broadband providers need to invest a portion of their profits in the bandwidth needed to meet consumer current and future demand. Today's heavy user with be the average user in a year or so.
Todd Spangler commented:
Paul, I meant that organic growth rates on the order of 50% annually in a relatively mature industry like cable are unheard of (aside from M&A activity).
Paul commented:
Todd, thank you for the updates! That this operator has not grown 50% YOY is distinctly different than "would not have been able to" though. It feels like the implication is that it's "not possible" for an operator to grow 50% annually. Starting with very small numbers though, the percentage of growth year over year can skew drastically. The smaller the subscriber count, the larger the percentage of growth that gets represented by minimal additions. Along those lines and without knowing the numbers (or anything about Mid Continent) beforehand, it isn't completely unreasonable to start with 50,000 Internet subs in 2007 and grow 50% YOY, while still resulting in the 250,000 ballpark provided for mid-2011. That Mid Continent has grown at a much slower pace than I thought possible (without knowing anything except what was represented in the graph) doesn't mean they "would not have been able to".
Todd Spangler commented:
On the question of what the y-axis shows, Midcontinent says it is aggregate peak usage across circuits provisioned from multiple upstream providers.
Todd Spangler commented:
On the question about subscriber growth:
Midcontinent would not provide specific subscriber numbers but said residential data customers grew 9.7% from the end of 2009 to the end of 2010. In 2011 year to date residential data subs are up 1.9%.
Todd Spangler commented:
Regarding subscriber numbers: No, Midcontinent would not have been able to grow subscribers 50% YoY. The operator cited having “more than 240,000 customers” in June 2009; in July 2011, it says it serves “more than 250,000 customers.” I’ve asked for more detail from the company.
Bathswana commented:
What is that graph showing? Is the y-axis a percentage of capacity? Are they measuring the backbone or the last mile? "Internet utilization" is a very slippery term.
And the data points (which I assume are measured weekly) seem a little too uniform to reflect real world weekly consumer usage. They correspond to the red trending line almost exactly.
I say the graph is fake. Made up from nothing to prove a point.
Paul commented:
With regards to the Internet Utilization graph posted in this article, how does the growth in Internet Utilization relate to subscriber growth? If online/Internet penetration rate among homes passed grew proportionally (ie adding 50% or so year over year), then the usage "per customer" wouldn't look so severe even if the usage across the system appeared to increase exponentially with the system-wide growth. And the graph refers to mid-west, so maybe doubling the penetration rate 10% to 15% to 23% to 30% to 40% (for example) from 2006-present wouldn't seem so abnormal year over year.
Peter Litman commented:
The Comcast user's story is a distressing one. He's heavy user of the service and generally likes it. I'd think a company would want to find a way to keep such a customer -- who clearly values the service and is using it in a legal way -- and, if necessary, charge him more, rather than shut him off and not let him back for a year. That's not just draconian, it is also capricious.