The Gigabit Household: Only a Matter of Time

Any way you look at it, the telecommunications business remains dominated by one issue above all: bandwidth, and how to get enough of it to satisfy end users without breaking the business – and the bank.

For years now, the Internet's not-quite-believers have been saying that demand for Internet bandwidth was finite and would eventually start slowing down. As the market saturated and consumers ran out of things to do with all that capacity, demand would tail off. Yet there's no sign whatsoever that this is about to happen.

Look at the most recent traffic data from the Amsterdam Internet Exchange B.V. (AMS-IX) , the world's largest Internet traffic exchange point. The volume of accumulated traffic handled at the exchange grew by about 54 percent between June and December 2008, and the trend continued strongly in January 2009, with a whopping 10-terabyte rise last month in traffic handled.

Increases of that order are consistent with a very long-term trend that has seen Internet bandwidth demand rising by around 80 percent a year, year after year, for well over a decade, regardless of the number of applications, type of applications, type of end users, type of end user devices, and other metrics that might be expected to disrupt the trend line.

There is absolutely no reason to suppose that this steady year-on-year rise in bandwidth consumed will not continue for as far out as we can see. One recent contributor to that rise, for example, is the very rapid spread of mobile broadband and devices attached to it, as described in a column last week by my colleague Patrick Donegan. (See Verizon Rides a Mobile Data Wave.) In the future, a whole raft of new consumer service ideas such as 3D TV and gaming, consumer telepresence, ultra-high-definition video, consumer and business cloud computing, and holographic video are all in commercial development or in the lab.

However, the exact nature of the apps, end users, or devices that will ratchet up demand doesn't really matter that much to those planning network requirements. The consistent history of 20 years, along with the miraculously flexible character of the Internet, means it will continue to generate new ideas at a steady rate. Even though we don't know what those ideas will be, telcos can safely assume that bandwidth consumption will continue to grow in line with historic precedent.

In practice, that means – for example – that it's fair to assume that the average household will be running a fixed Internet connection at about 1 Gbit/s 10 years from now. This therefore must be the objective that everyone aims for.

Discussions about Internet bandwidth inevitably have a chicken-and-egg character: If telcos don't make bandwidth available, demand will automatically be tamped down. But that would be a pretty dangerous game to play, alienating end users, regulators, and politicians, and ultimately backfiring on those who are lagging the global trend. The long-term requirement is clear. The only question – not a small one – is how to get there.

— Graham Finnie, Chief Analyst, Heavy Reading

Page 1 / 2   >   >>
optodoofus 12/5/2012 | 4:12:28 PM
re: The Gigabit Household: Only a Matter of Time I don't believe this analysis is valid, and I think you have made an unfounded logical leap here.

The Amsterdam Internet Exchange traffic may be growing briskly, but that does not mean that the demand of indivudal subscribers is growing at the same rate. The aggregate bandwidth demand could grow if indivudal bandwidth consumption was constant (or even declining) - as long as there was robust subscriber growth. Your analysis concludes that it is just a matter of time before everyone needs 1Gbps connections to their house. However, I do not believe the data you cited supports this conclusion.

I found this analysis of traffic demand growth in Japan to be very interesting.


Per-user bandwidth growth in Japan is estimated to be 18% - much less than the 80% bandwidth growth you cite in the article. To really understand the future demands of an individual household, you cannot just base it on the aggregate demand and ignore the subscriber growth. When you adjust for subscriber growth, the picture looks very different.

I have a mid-tier FIOS connection in my home now, and it provides more bandwidth than we need. It is used for video-on-demand service, and we consume music services and over-the-top video as well. And what we have today meets all our needs. So, why should I expect that we eventually need a 50-fold increase in capacity to my home?

I keep hearing this refrain over and over again. But just because people keep repeating it doesn't make it true

scfm 12/5/2012 | 4:12:27 PM
re: The Gigabit Household: Only a Matter of Time ------
I don't believe this analysis is valid, and I think you have made an unfounded logical leap here.

(lots of good stuff deleted)

I keep hearing this refrain over and over again. But just because people keep repeating it doesn't make it true

Well said. I had the same reaction to the article, but didn't have any facts at hand on which to base a rebuttal.

At some point, increased bandwidth will have to be justified by significantly increased revenue, not just subscriber retention in the face of competition.

AT&T and Verizon are spending truckloads of money putting in the infrastructure to get homes to the 20-50Mb/s level and, as you illustrate, that is more than most can even consume. They are going to need to operate these networks at these speeds for many years to pay back the investment.

I don't see applications being developed any time soon that will motivate consumers to direct a sufficiently higher portion of their disposable income to communications to justify a carrier business plan to spend even bigger truckloads of money chasing 1Gb/s to the end user.
paolo.franzoi 12/5/2012 | 4:12:27 PM
re: The Gigabit Household: Only a Matter of Time
At some point, 1 Gb/s per home might be real. Are we not all arguing (under the heading of Net Neutrality) that USING our existing 1.5Mb/s connections all the time (like having P2P enabled) is a problem?

So, the 80% increase could stem not just from increased subscriber count but also greater usage of existing facilities.

Larry, Monkey 12/5/2012 | 4:12:26 PM
re: The Gigabit Household: Only a Matter of Time But have you ever seen an invalid leap? It's a sight you won't soon forget!
fiber_r_us 12/5/2012 | 4:12:25 PM
re: The Gigabit Household: Only a Matter of Time The article seems to broadly state that in 10-years time that households will require 1Gb/s of bandwidth to support future applications. You seem to argue the minor point of the data that was used to come to that conclusion (looking at growth of traffic at an exchange). However, the conclusion is probably correct - unless you are one of those that believe "no one will ever need that much bandwidth" - in which case you can get in line with all of those who have made similar capacity statements for practically every other technology and been subsequently proven to look quite foolish in retrospect.

Consider: 10 years in the past most households were on 56Kb/s dial (or less), and no one envisioned the various apps we all use today. Today, almost everyone can get single-digit Mb/s (with some lucky minority receiving double-digit Mb/s). This is a 100 to 1000 fold increase over the last 10 years.

Similar growth of the next 10 years would place even the low-end users in the multi-hundred Mb/s range. Why wouldn't you expect a similar growth over the next 10 years?

Technologically, we can easily deliver 1Gb/s to the homes today. It just costs a lot to put the fibre in place. But, no more so than other past projects to put in the old copper or coax plants.

As the article states - it is only a matter of time.
Hanover_Fist 12/5/2012 | 4:12:22 PM
re: The Gigabit Household: Only a Matter of Time kinda like watching Bob who has no arms or legs swim...right?
fiber_r_us 12/5/2012 | 4:12:22 PM
re: The Gigabit Household: Only a Matter of Time "Bob" seemed to make pretty incredible progress over the last ten years!

Really - how about trying to contribute to the discussion!
hyperunner 12/5/2012 | 4:12:20 PM
re: The Gigabit Household: Only a Matter of Time OK, so I know I might be a bit traditional here, but who will pay for the cost of upgrading the access and core infrastructure to support 1G service to the customer - even if a suitable demand for this service exists?

Even with huge over-booking at the packet layer this is a massive cost. And given the nature of emerging high-demand application I can tell you that over-booking ratios have to come down, not go up.

Here's a suggestion for Light Reading: ask the core router vendors, and all the DWDM vendors for a historical "price per Mbit per second" of their products.

How much did 1Mbit/s of core router capacity cost in 1995, 200, 2005, and what's the forecast price in 2010?

This is only part of the "cost price" of broadband service to the service provider. Another big chunk is the cost of operating all this IP stuff that was never designed as a service provider technology - but that's a complicated topic!

It would be interesting too see if the Ciscos and Cienas of this world have been able to lower their product prices for IP and DWDM capacity as fast as demand has increased.

My gut feeling is they have not done that. And yet over the same period we are told that the average US family will spend about the same amount on fixed telecom services.

Users don't care about bandwidth - they just want their video and their music to arrive quickly and reliably. If it takes a huge investment by the service provider to do that then so be it.

So where does the money come from? How can service providers offer a sustainable business?

fiber_r_us 12/5/2012 | 4:12:20 PM
re: The Gigabit Household: Only a Matter of Time Certainly over the last decade the cost/Gb of router capacity and transport capacity has dropped precipitously. I would argue it cost no more today to upgrade the infrastructure than it did 10 years ago - yet you would easily get 10 to 100x the BW. Exact numbers would depend on your architecture, equipment choices, etc.

Example, in the late '90s OC192 router cards were in the $1M range at one point. Today, I can get 10G router ports sub-$10K - on routers that support over 100x10G ports. Evolution in silicon will allow this to continue for some time (routers mostly still use 90nm silicon for example).

Another example, nx10G DWDM systems prices have also plummeted over the last few years. Much easier to build massive WDM capacity in metro areas (if you are the carrier with the fibre). And, 100G wavelengths over metro-type distances should be fairly achievable in the near future at reasonable cost points.

Between access kit evolution, router evolution, a transmission evolution, it is fairly easy to visualise how to build an infrastructure supporting 1G access speeds at a reasonable price point. The majority of the cost is in laying the last mile fibre - not in the routers and transmission gear.
paolo.franzoi 12/5/2012 | 4:12:19 PM
re: The Gigabit Household: Only a Matter of Time
I think you are missing the point guys.

I 1 Gb/s access rate does not mean a guaranteed 1 Gb/s rate to much beyond the first interface. So, the costs are astronomical. I try to use this example: Imagine a 10,000 dwelling city (that is about 50,000 people or less) - each requiring 1Mb/s. That means that town (which is quite small) now needs 10Gb/s transport. Okay now start scaling the bandwidth.

So, there have to be changes to accomdate all of this. These mostly have to do with caching and storage because we are not running 10Tb/s to a small city of 50,000 in 10 years. If we do this caching and storage, I think the need for increased bandwidth will drop dramatically. What will increase dramatically is the average machine to machine communications with a dwelling (for example downloading today's TV shows overnight).

Page 1 / 2   >   >>
Sign In