The exponential growth of internet traffic spurred by bandwidth-hungry applications such as YouTube poses problems for telcos.

IT HAD to happen. I've now seen the word "zettabyte" in print. A zettabyte is 1000 exabytes, which is 1000 petabytes, which is 1000terabytes, which is 1000gigabytes.

In other words, it's a big number. Really big, as Douglas Adams used to say.

Not only was the word used in print, it was used in the title of a report, Approaching the Zettabyte Era, from networking company Cisco. If you're at all interested in the future of the internet, I recommend this document to you. Search for it on Cisco's website or go to tinyurl.com/4wucvf.

It details the exponential growth of internet traffic. In the nature of exponential growth, the increases will be much more substantial in the future than they have been up to now.

The report outlines the quantity and quality of the increase.

The big change will be a massive increase in video as a proportion of traffic on the net over the next few years.

A few snippets:

Global IP traffic will nearly double every two years until 2012, when it will exceed half a zettabyte. That means the 2012 internet will be 75times larger than it was in 2002.

Traffic will generate 28 exabytes a month in 2012, equal to 7billion DVDs.

Internet video is about a quarter of consumer traffic but will hit nearly 90% by 2012. It would take well over half a million years to watch it all.

YouTube is just the beginning. Internet video growth is in its infancy. Internet video to the PC screen will soon be exceeded by a second wave arising from the delivery of video to the TV screen. After 2015, a third wave of video traffic will result from video communications.

Despite this, the internet is not collapsing under the weight of online video. Service providers are accelerating their infrastructure upgrade plans in response to the growth.

Mobile data traffic will double each year from now until 2012. Broadband-enabled notebooks are creating sharp increases in mobile traffic. In some parts of the world, mobile broadband is a substitute for fixed broadband. The result is that mobile data traffic in 2012 will be 20 times greater than today.

There's much more but you get the picture. Massive increases in traffic, so much so that today's speeds, volumes and costs will seem laughable in just a few years. It's interesting that Cisco says that internet infrastructure will keep pace.

Not everyone shares this optimism. Such volumes cannot be reached, let alone sustained, without vast improvements in bandwidth at the backbone level and to the desktop.

The applications that Cisco foresees, which will include high-definition video-on-demand, will place great strains on infrastructure. Cisco's report talks of "surprises" or spikes in demand that will exceed the network's capacity to cope.

There are many contrary views. Some point out that Cisco, being the largest supplier of the products that make the internet run, has a vested interest to talk up its growth. That may well be the case but it does not refute the argument.

Another recent report, from David Payne at Swansea University in Wales, agrees with the Cisco predictions on bandwidth growth and on the fact that it's driven by video but is much more pessimistic about the infrastructure's ability to grow to meet these demands.

Dr Payne makes the point that it's not possible without the universal introduction of FTTP (fibre to the premises, otherwise known as fibre to the home, or FTTH).

FTTP is necessary but not sufficient.

Also needed is substantial investment in the backbone infrastructure by network operators, who need economic incentive.

Then there's the contrarian views of Andrew Odlyzko, from the University of Minnesota, who has long argued that content is not king and that internet growth rates are slowing. He cites the example of Hong Kong, where bandwidth capacity has outpaced traffic growth and where operators have trouble making money.

These disparate views are not too far from each other. Growth rates are exponential - the main point of divergence is the exponential growth factor and disagreements over whether content drives bandwidth or bandwidth drives content.

I don't reckon it really matters. It's just a matter of degree. No matter what happens there will be enormous changes, and internet volumes will increase to the zettabyte level in not too many years.

It's only a few years since the International Standards Organisation devised "zetta" as the next multiple of a thousand beyond "exa". It had to do it because of the aforementioned exponential growth in computing and communications volumes.

They took the precaution of adding an extra one - "yotta" - beyond zetta. There's a lotta bytes in a yottabyte.

graeme@philipson.info