(这条文章已经被阅读了 227 次) 时间：2001-01-19 09:58:22 来源：sos遥然 (sos) 转载
The P2P Report
By Emelie Rutherford
What is P2P?
Is P2P computing new?
Who’s leading the pack?
What are the three flavors of P2P?
How does P2P fit into the enterprise?
What problems do CIOs have with P2P?
How can P2P help my company?
What is the future of P2P?
Despite the NASDAQ downturn, a new technology trend is making its
way into pundits and technology vendors’ hearts: peer-to-peer (P2P) computing. The Gartner Group says P2P will “radically change business models.” Andy Grove, chairman of Intel, calls P2P “a revolution that will change computing as we know it.” And vendors such as Lotus, IBM and Hewlett-Packard are rushing alongside Intel to build P2P applications. Yet this type of computing and its repercussions on the IT community are still not understood by many people.
What is P2P?
P2P is often described as a type of decentralized computing where computers communicate directly with each other. A more common (and limiting) definition says that P2P environments consist of computers with equal capabilities that share resources (such as processing power and memory), communicate exclusively with each other and do not connect to servers or central databases. Some P2P setups and applications, however, stray from this purist definition and include servers in the mix.
P2P computing environments pool together processing power and memory from many computers. A well-known organization that uses P2P is SETI@home, a group that probes the universe for extraterrestrial intelligence by analyzing signals received from telescopes, all with the aid of the combined processing power of millions of computers, whose owners install SETI@home’s software.
P2P applications — such as Napster, a website that provides software with which people can search for and swap music files on each other’s computers — allow users to communicate synchronously, doing things such as instant messaging, working on shared documents and searching each other’s computers for files.
Essentially, P2P allows computers, and their users, to tap unused resources — such as extra megahertz of processing power, gigabytes of storage and the rare music gems — that would otherwise remain locked up in individual desktops.
As more vendors and CIOs try to make sense of this burgeoning space, there is more discussion about the definition of P2P. Some create what they consider to be P2P computing environments with servers. They regard P2P as simply the efficient use of networked computers that do not have to be equal, and can scale and allow a more powerful computer to assume the bulk of processing demands. In fact, some applications, including Groove, require a server-based relay service for when users want to make their workspaces accessible to people outside their firewall.
Is P2P computing new?
Despite the flurry of recent attention generated by Napster, (which was sued by the recording industry in 2000 for violation of copyrights), P2P computing has been around for several decades. But alternate forms of computing, such as client/server, have been more popular, in part because they allow CIOs to have central control of networks and because they are cost-effective (computers that act as terminals are cheaper than those with their own memory, processing power and applications).
It wasn’t until the speed of PCs increased (as well as the number of them connected to each other via the Internet and private networks) that P2P computing really picked up. One of the earliest large-scale uses of P2P was in 1994 (when fast processors, like Intel’s Pentium II, were on the market), when two scientists at the Goddard space Flight Center in Maryland networked 16 processors together and created a single cluster computer. In 1999 Napster launched and, a few months later, SETI@home began distributing its P2P software.
These two efforts received a good deal of press in 2000 and helped spur the current craze. Intel Capital, the chip maker’s venture capital arm, estimates that over 80 vendors are trying to make their way with P2P applications, though most have not yet launched. Many seem to be building applications focused on increasing the exchange of intellectual property in the enterprise.
Who’s leading the pack?
One P2P vendor that is receiving a lot of attention, in part because it was started by Ray Ozzie, the creator of Lotus Notes, is Groove Networks. Its application, called Groove, (which, as of this writing, is still in its beta stage) is a type of operating system that lets users create virtual work spaces in which they put files that others can browse and use. They swap and work on shared documents from each other’s PCs, through a browser-based interface, as though they were central databases.
Another company receiving a lot of ink is InfraSearch, a soon-to-be-launched search engine that will reportedly return images, solutions to calculation and other non-text data that cannot be found by engines like Yahoo that scour only their own indexes instead of external databases. This Calif.-based startup is reportedly planning on setting up private networks for businesses so that users can search each others’ computers for hidden documents or supply chain partners’ computers for real-time prices, inventory status, or catalog listings.
What are the three flavors of P2P?
There are three basic divisions in P2P land: collaborative computing, instant messaging, and affinity communities.
COLLABORATIVE COMPUTING environments (also called distributed computing, shared processing, cycle sharing, and multi-agent systems) pool the processing power of many computers. SETI@home is collaborative computing’s poster child. Some vendors like Popular Power, United Devices and Entropia recruit people from the general public to volunteer their computers’ idle processing power. In return these netziens either receive incentives or the satisfaction of knowing they are helping a worthy cause (many recipients of the processing power are non-profit organizations). Collaborative computing environments are popular with science, biotech and financial services firms that need intense computer processing. Intel, for example, has reportedly been using P2P computing since 1990 to run broadband-hogging chip design applications. But many enterprises that don’t have such demands for processing power, have not embraced collaborative computing because such environments are very difficult to manage and require much coordination.
INSTANT MESSAGING applications allow users to swap text messages synchronously. Commercial vendors such as AOL and Microsoft offer free messaging tools. Most enterprise vendors are simply trying to duplicate the success of these apps while tailoring them for enterprise use.
AFFINITY COMMUNITIES are direct file sharing groups, like Napster. They use shared spaces in which they interact directly without dealing with servers and boundaries, doing things such as collaborating on documents in a shared space, instant messaging each other, and searching each others’ computers. Vendors in this arena include the aforementioned Groove Networks, GoneSilent, Pointera, Roku, uRoam, Hilgraeve, FlyCode, Hotline Communications, Kalepa Networks and Centrata.
Some industry watchers predict that some commercial e-commerce ventures will start using P2P technology. On the heels of Napster, these companies will set up systems in which groups share data, but pay, for example, to download a peer’s e-book file. People could also sell items listed on their computer in an auction manner without dealing with an auction site like eBay. Other P2P observers have pointed out similar uses of P2P for online marketplaces and for companies who want to extend their content network to customers, as an added value.
How does P2P fit into the enterprise?
P2P proponents say that businesses can save billions by using distributed computing setups that take advantage of unused bandwidth and resources. Messaging tools and affinity communities can open up intellectual property and data that are otherwise hidden in departmental offices and servers. Now that knowledge management is such a big priority for so many companies (80 percent of the world’s biggest are dabbling in KM, according to PricewaterhouseCoopers and the Conference Board), businesses can reap benefits from P2P that they don’t care to measure in dollars.
What problems do CIOs have with P2P?
Security remains one of the biggest issues. Because P2P is ad hoc and decentralized in nature, IT departments can’t have the same level of centralized control they have with client/server setups. This is where some P2P users have introduced servers, so they can hold on to some central control, as well as provide security for when users of messaging and affinity communities can contact people outside the firewall. CIOs may want to consider further security precautions, such as watermarking, which allows for authentication of a given piece of software.
In addition to security issues, P2P setups also place a huge burden on individual computers. The applications can chew through a lot of bandwidth and even, inadvertently, reroute the flow of network traffic. Further, applications do not always work together in P2P environments.
How CIOs work around issues of security, scalability and oversight remains to be seen. Few organizations (with exceptions like Intel) have established and are talking about their own P2P networks. And few vendors have proven themselves, though they are touting solutions to some of the aforementioned issues.
Many messaging and community vendors say they built their solutions with encryption and algorithms that safely let users configure their setups. Those dabbling in the network realm are proposing more scaleable solutions that allow lesser computers in P2P networks to take a subservient role to the more powerful computers. But there’s no proof yet.
One thing, however, is near certain: If a P2P network or application becomes popular in an organization, CIOs will likely have to beef up the computers’ bandwidth and networks’ capacity. (Case in point: Indiana University banned file-swapping programs when it discovered Napster was consuming half of the university’s network.)
Despite the ambiguity and threats such P2P solutions pose, CIOs should not dismiss P2P. Their company’s users are most likely already using chat tools like AOL’s instant messenger to conduct work, thus bypassing their security infrastructures.
How can P2P help my company?
Unless you work for a company that requires a lot of computing power to complete a task (such as a biotech, financial services, chip manufacturing firm, or an organization focused on breaking codes), you likely will not be installing a collaborative computing environment.
But, across the board, enterprises can theoretically benefit from P2P messaging tools and affinity communities. The messaging tools allow synchronous communication that is impossible with email. So remote workers collaborating on a project can instantaneously chat and complete a task.
The affinity communities will enable companies to tap intellectual property locked up in boundaries such as separate offices and servers with restricted access. If they’re opened up, employees can save time by instantly search for documents. (Imagine a lawyer looking for a legal document, written by a coworker for a different client, that he knows he could quickly modify for his present client, but can’t find anywhere on his network. If he’s in a P2P community with that other lawyer, he could search his computer for the document and quickly retrieve it.)
What is the future of P2P?
Aside from the furious attention P2P has received from the press and new vendors, there have been few attempts to define protocols for it. Intel seems to be positioning itself to be at the forefront of such nascent activity and, in August, 2000, formed a consortium called the Peer-to-Peer Working Group that invites anyone to help it debate and develop such tenets. The Foundation of Intelligent Physical Agents, a non-profit standards body, has also formed a group to create and propose actual P2P standards.
One way to stay on top of this morphing space is to join the P2P discussion mailing list run by the Peer-to-Peer Working Group. Members in the discussion thread actively debate the definition of P2P computing and applications, touching on issues mentioned in this article. (To sign up, visit the members page at http://www.peer-to-peerwg.org/members/ and follow instructions for registering under the heading “Mail Forum.”)
Emelie Rutherford, Web Writer, can be reached at [email protected].