Intel’s Chief Technology Officer Pat Gelsinger has warned that the Internet is becoming dangerously overloaded and could be close to collapse. In short, there’s a capacity problem. As millions of new computer users from developing nations begin to get connected, the thirty year-old Internet architecture is struggling to keep up. Consider this: the current IP version 4 (IPv4) protocol was standardised in 1978 – that was 26 years ago.

But this is only the half of it. While the number of worldwide Internet users is currently estimated to be in excess of 800 million (click here for more details), Gelsinger told the Intel Developer Forum (IDF) that the “the number of devices on the Internet will grow into the tens of billions.” The slow but gradual implementation of IPv6 will help to alleviate some of the strain, but it’s a stop-gap measure. It doesn’t address the high-capacity demands of VoIP telephony and HDTV broadcasts over the web, nor does it tackle existing security and reliability issues. What is certain, though, is that both users and future web-based applications will demand much more bandwidth than the current infrastructure currently delivers.

Time for the Internet version 2?
So what Gelsinger has proposed is a newer, safer and faster Internet. And Intel just happens to have one waiting in the wings. Crucially, its PlanetLab initiative isn’t a replacement for the Internet. With so much revenue and critical data relying on the current architecture, it would be impossible to dump the ageing TCP/IP protocols completely. The PlanetLab project works by using the existing Internet infrastructure, but it adds a global network of Intel computers on top to attach extra functionality to it.

The goal is to create a 1,000-node network, an intelligent platform that can host a variety of smart web and grid computing services. Today’s Internet services are linked to local servers; if a server goes down, you lose the service. The PlanetLab approach shares services across the global network, so if a server goes down, the service remains up and running, as the load is dynamically re-routed to other PlanetLab nodes.

“When I get my network service from an ISP,” says Gelsinger, “I’m essentially renting a slice of a whole bunch of Cisco routers. In the future, when I get a PlanetLab service, I’ll be getting a slice – a virtual machine of a whole bunch of PlanetLab servers that are built into the network. [Such] a planetary-scale overlay of computational services would open the Internet up to a new era of innovation while complementing other Internet initiatives. The network would have a global view of the network itself.”

It’s alive! It’s alive!
The idea of a ‘self-aware’ Internet is arguably the most interesting element of PlanetLab. Intel, for example, has been developing a service to improve the overall security and efficiency of the Internet. Dubbed PHI (Public Health for the Internet), PlanetLab users monitor the condition of the network and share the data among each other. As a result, as Gelsinger explains, the network can “see the transactions, the trends, the different attacks and services and viruses on the Internet and begin to self-respond.”

At IDF, Gelsinger demonstrated how viruses could be identified and tracked across the PlanetLab network. PHI is based on what Intel calls PIER (Peer-to-Peer Information Exchange and Retrieval). This is a massively distributed query system, a software tool that measures the health of the Internet and provides connected machines with info about the latest worms, viruses, and the IP addresses of machines that are attacking other machines in large numbers. Each computer running the P2P tool, therefore, contributes to an overall picture of the Internet’s status.

“Imagine that we took PHI,” says Gelsinger, “and said we’re going to notify the firewalls of corporations of the IP addresses for the top ten current attackers on an ongoing basis. And they’d be dynamically updating their firewalls to just block those top ten. [We estimate that] 60 per cent of service attacks would be eliminated before they ever hit the corporate network.”

Whatever way you look at it, the web needs an overhaul. “The Internet has some rigid technical aspects, which make it slow and hard to implement improvements,” adds Joerg Hennig, VP Networks and Datacentre at web host 1& 1 Internet. “The Internet itself is built on a layered structure, so the idea of building an overlay to the Internet appears to be a practical way to make enhancements. However, with dynamic content in particular, there are still a lot of unanswered questions and challenges to assure a consistent distribution.