There is a way of connecting computers together without a server called peer to peer where each machine can gain access to the resources of all the other machines.
Now this concept has been extended to the Internet allowing it to act as the network to link machines, blowing away the need to physically tie these computers together.
An example of this Internet based connectivity is Napster, the Web operation that has so angered copyright people around the world. With Napster computers are connected together with a limited resource - namely any MP3 digital sound file on a hard disc - made accessible to any other Napster user anywhere in the world.
So you end up with a huge network of users and hard discs with hundreds of thousands of files available.
Another example of this type of computing, and perhaps a more relevant one to the future of Internet-based computing is Project SERENDIP, which the SETI Institute is a major supporter of. These scientists are looking for extra terrestrial intelligent life forms and have built a system called Seti @ Home which ties together a lot of machines via the Internet, downloads a chunk of data to them, data pulled from its Arecibo Radio Telescope, in Puerto Rico.
What happens is that loads of machines, and this figure has hit 1.6m before now, crunch data collected from the telescope. Think about it, 1.6m processors working on your behalf. Not a bad idea at all. The SETI Institute then stitches this data back together and the whole thing moves along looking for life beyond our planet.
So we have this giant supercomputer that is built up from many distributed nodes - or PCs connected to the Internet - which do all the hard work. So does this type of computing model have a place in the business world?
P2P is a brilliant idea and one that will form the backbone for the future of the Internet but not for a while yet. This idea of creating supercomputer number-crunching engines - and this is the basis of what we are talking about here, bloody big abacuses - will impact the consumer world first. This is the likes of SETI benefiting from the generous donation of processing power from people eager to find out if life exists on other planets.
But down the road here we have a business impact as well. Think about a truly connected Internet that has continuous broadband connectivity -- where 'spare' processing power can be put to good use.
Now imagine that in the future this power can be paid for and not simply given for free as it is now. Maybe this is not payment in cash, but maybe payment in free access or other virtual services. But payment in some form is made. After all, the world is not populated by charities.
From a business perspective then having access to this processing power is quite exciting. Imagine a company involved in genome research, a complex subject that takes a lot of number crunching. Maybe that company wants access to a supercomputer, say something like IBM's ASCI White, and it needs TeraFLOPS of computing power. Clearly that amounting of computing power does not come cheap.
So in rolls the SETI style operation. Now SETI on its Web site claims that IBM's ASCI White operates at 12 TeraFLOPS and costs $110 million. It adds that its Internet-based solution crunches data at about 15 TeraFLOPs and has cost $500,000 so far.
What we are talking about here is using the dormant power of the Internet to circumvent the historical large investments that had to be made to acquire supercomputer class power.
The logic is a simple one: divide up giant computing tasks into manageable bits and split these pieces out across tens of thousands of computers connected to the Internet. You've just built a massive parallel computer.
All you have to figure out is security, access control, fees, and other management issues. The concept, or rather the idea, is incredibly exciting and will help turn the Internet into something much more tangible than simply a giant network of networks.
Back to Share and Share alike
This was first published in July 2001