Showing posts with label IP networks. Show all posts
Showing posts with label IP networks. Show all posts

Sunday, August 19, 2007

Cisco Predicts Exabyte Networks

Cisco's recent forecast of global IP bandwidth consumption suggests a 37 percent cumulative average growth rate between 2006 and 2011, or about five times the 2006 level. That's aggressive, but you might expect that. You might even have expected the prediction that consumer usage will outstrip business usage, though business dominates at the moment. You wouldn't be surprised at all to learn that video will drive overall global usage.

You wouldn't necessarily be surprised to learn that Cisco forecasts at least 60 percent of all traffic will be commercial video delivered in the form of walled garden services. And a significant percentage of the remaining 40 percent of IP bandwidth will be consumed by IP-based video applications.

The next network, in other words, will be a video network that also carries voice and non-real-time data.

That would be a stunning change from the originally envisioned view of the Internet. But I think we have to recognize at this point that virtually none of the key developments in communications technology have developed as industry insiders, public policy proponents, technologists or entrepreneurs had supposed.

To be sure, all of the diligent work on Session Initiation Protocol will have a significant payoff. But that didn't stop Skype by rocketing past SIP using a proprietary approach.

The Telecommunications Act of 1996 was supposed to lead to an explosion of innovation by dismantling restrictions on "who" could be a provider of Class 5 switch services. Instead, innovation came from the Web. Perhaps despite the Telecom Act, all sorts of innovation has happened.

VoIP was supposed to transform the nature of communications. Instead, mobility, instant messaging and social networks are doing so. One might arguably look to all manner of text communications as the disruptive communications development of the past several decades, not voice.

And then there's electronic numbering and voice peering. Perhaps these approaches still will have some dramatic impact on global voice communications prices and ability to circumvent the "public network." But it's starting to look as though ENUM might be a next generation to provide the signaling system 7 function. That's not to say it is unimportant: only to say it was not what many had intended or expected.

So far, it would seem that the most disruptive impact of the whole basket of new technologies has been to disrupt our ability to predict the future. We've been wrong more than right, as we always are. IP networks are not now, and never will be, as closed as the old public network was. Neither are IP networks going to be "open," any-to-any networks in the old manner, with no intelligence or policies operating in the core of the network.

Lots of things can, and should, be done "at the edge." But increasingly, lots of things cannot. The transition of the global IP network to video also means a shift to real time services (and we aren't even talking about the same process at work for voice and visual collaboration). That spells the end of the completely "dumb network."

"Tokens" are the New "FLOPS," "MIPS" or "Gbps"

Modern computing has some virtually-universal reference metrics. For Gemini 1.5 and other large language models, tokens are a basic measure...