Like the Web services market, the P2P sector is embryonic but, as it grows, it is likely to concentrate on to two main areas: file sharing and process sharing.
File sharing will bring Napster-like capabilities to the corporate space, enabling different users to share files directly from their machines without needing to go through a server.
Process sharing involves distributed resource management, brokering spare CPU time on client machines across the network to complete tasks more quickly and make more efficient use of computing resources.
As it is a low-level communications framework for P2P applications, Jxta covers both of these areas. The system doesn't really care what it makes available to its peers - instead, that is left up to the application itself.
Participants in the P2P network discover each other by either broadcasting over a subnet, inviting another peer onboard, examining another computer's available connections, or by using a single peer as a rendezvous point.
Peers can advertise their content or services using an XML-based "advertisement" with a data footprint of roughly 5Kbytes. When a client wants to make a file request, it sends a request message via the proxy to access the content.
Jxta works across firewalls by using HTTP proxy servers outside the firewall, connecting to a single client in a peer group inside the firewall. This client can then communicate requests and advertisements to the rest of the peer group.
Communication channels are defined using unidirectional "pipes" that are bound to a peer at runtime using Jxta's Pipe Binding Protocol. Pipes can be point-to-point connections or one-to-many communications channels. They can also be moved between peers for fault tolerance purposes.
The whole system is designed to be transport-independent, meaning that it can run over IP or non-IP services, such as IrDA infrared connections for personal digital assistants.
Sun's motives for releasing this technology are questionable. On the one hand, the technology has been developed under an open source Apache licence, and the source code is freely available on the Internet. On the other, Sun clearly has an agenda, as it is already developing applications for content sharing using the technology, and there are no plans yet to submit Jxta to a standards committee.
Sun strategist Gene Kan explained, "Right now, we are in the very early stages of developing Jxta and it doesn't make sense to go that route." He added that the company wants the industry to standardise on the technology and proposed a strategy of "standardisation through adoption".
The outcome of this strategy remains to be seen, but Kan said Sun has so far failed to negotiate with existing industry bodies such as the New Productivity Initiative. The New Productivity Initiative has unveiled a distributed resource management specification with which there is bound to be some overlap.
Meanwhile, Microsoft's Systems and Networking Research group has been quietly working on its Farsite system, a P2P file-sharing technology designed to create a single, huge logical drive from many physical drives across the network.
The Farsite project is pure research at the moment but, once the marketing drones get hold of it, it should not be too long before the technology behind it is packaged into some sort of Jxta spoiler.
Getting Wired, p56
The benefits of P2P
This will bring Napster-like capabilities to the corporate space, enabling different users to share files directly from their machines without needing to go through a server
This involves brokering spare CPU time on client machines across the network to complete tasks more quickly and make more efficient use of computing resources.
This was first published in May 2001