Trend number one: Computing
December 04, 2000
by Om Malik
Issue: 86, Page: 95
Distributed computing redefines computer networks, underpinning innovation, company formation, and investments.
Ignore the hype proclaiming Napster-style peer-to-peer computing as the greatest thing since sliced bread. In reality, peer-to-peer, or P2P, computing is just one-third -- grid computing and distributed information infrastructure being the other two-thirds -- of a very hot idea: distributed computing.
In 2001, distributed computing will become one of the Sand Hill Road crowd's sure things. Signs of this are already evident, with venture capitalists throwing hundreds of millions of dollars at companies like AppleSoup (now FlyCode), Centrata, Quiq, NextPage, Distributed Science, Engenia Software, Popular Power, Static Online, United Devices, and Uprizer.
But there is more than money at stake here -- distributed computing is going to change the way we think about technology and the Internet itself. On a practical level, it will have far-reaching effects on how software is written, and on how computer and networking gear is installed in businesses.
Within the framework of distributed computing lie two other technologies that are even more promising than P2P -- grid computing and distributed information technology infrastructure. This year, companies that focus on these two components will get off to a blazing start, buoyed by the hype surrounding P2P.
P2P computing enables the direct exchange of services or data between computers. In this environment, the servers, desktops, and notebook PCs that make up a network become peers that contribute all or part of their resources -- such as processing power or storage -- to the overall computing effort. This type of architecture transforms client computers from mere consumers of services to service providers as well.
In a keynote at the Intel (Nasdaq: INTC) Developer Forum in August, Pat Gelsinger, vice president and chief technology officer of Intel Architecture Group, outlined how P2P computing could help corporations use their computing resources more efficiently. "Peer-to-peer computing could be as important to the Internet's future as the Web browser was to its past," he said.
P2P will spawn many companies that use Napster-like services to distribute video, music, files, and even software. But despite the evangelism of VCs like Softbank Venture Capital's Bill Burnham, no company has yet figured out a viable business model based on this architecture. "While the most visible impact of this model has been in consumer environments, peer-to-peer computing has the potential to play a major role in business computing," Mr. Burnham says. "Corporations can tap into existing teraflops of performance, and terabytes of storage, to make today's applications more efficient and enable entirely new applications."
Napster provides a primitive example of what these new applications might be. The music-sharing service's central servers act like traffic cops, keeping important data on members and shared file lists and performing music searches. While Napster users store and download music using their own resources, key data elements are kept in-house, on Napster's central servers.
Napster's technology applied to the enterprise supply chain -- where key data is kept within the walls of a corporation but the rest of the chain is broken up among suppliers -- is one way companies might use distributed computing. So far, few businesses have cracked this code, but we expect VCs to focus on startups chasing this viable business model over the next six months. Already, InfraSearch, a startup offering a search engine based on P2P, has raised nearly $5 million in funding and lists Marc Andreessen and Mike Homer among its angel investors.
Adam Beberg is a 26-year-old computer scientist and founder of Mithral Communications and Design, based in Santa Clara, California, which makes distributed computing tools and protocols. Mr. Beberg argues that the power of distributed computing lies in harnessing idle MIPS (or wasted computing power) by tying them together and using the cumulative power to perform supercomputer-caliber tasks. This is called grid computing, and it has inspired companies like United Devices, backed by Softbank, and Parabon Computation, which has raised $4.9 million in angel funding.
One person who predicts strong commercial demand for these kinds of services is David Kopans, chief financial officer of Applied MetaComputing. His company is already doing business with Boeing (NYSE: BA) as well as various educational and government institutions, and has received about $7 million in funding. The company's grid computing system, called Legion, promises to reduce the computing requirements from servers, cut storage costs, and use network bandwidth more efficiently.
Like Legion, distributed computing projects such as Globus, Beowulf, and Cosm (see below: Distribute the Wealth) will help accelerate the acceptance of this technology -- any industry that requires massive amounts of data crunching can benefit from distributed computing. Weather forecasting, feature film animation, protein modeling, and genome-related research are obvious fits for distributed computing. For example, a pharmaceutical company searching for a new drug by generating computer models of viral agents will find grid computing a cheap alternative to buying a supercomputer.
At Mithral, Mr. Beberg is currently working on the Cosm Project, a set of programming tools and protocols that will allow organizations to create their own distributed computing environments. Cosm's tools will help develop services that might make computer-processing power a commodity traded on the open market. Companies would be able to instantly purchase extra processing power from online exchanges like Porivo Technologies, one of a handful of grid computing exchange companies that are creating a portal that would allow commercial users to purchase idle computer power with one click.
While P2P computing and grid computing try to put down roots, the third element of distributed computing -- distributed information technology infrastructure -- is gaining momentum. New technologies like appliance-specific servers, Web switching, and content-distribution networks are proliferating. In the field of application-specific servers, Sun Microsystems (Nasdaq: SUNW) acquired Cobalt Networks (Nasdaq: COBT) for a cool $2 billion. Meanwhile, content-distribution networks from caching companies like Inktomi (Nasdaq: INKT) and Akamai Technologies (Nasdaq: AKAM) are another step in this direction. And Ejasent, a startup that launched its virtual applications and Web-infrastructure-on-demand services in October, has joined in to up the ante for distributed infrastructure. Taken together, these represent a distributed hardware and software architecture that shares a common bond: Internet protocol.
One of the biggest changes brought about by distributed computing will be a reconsideration of traditional network topologies and computing architectures. As the network infrastructure gets broken down into smaller, application-specific parts, reliance on larger servers will decrease. Software will become intelligent and tailored to specific tasks. This will allow companies to buy only the functionality they need.
Most important, though, distributed computing will help us realize much of the Internet's potential, transforming it from a document-based network of Web pages and email into a dynamic, granular network where specific components of information can be located and shared more efficiently. Finally, the network really is the computer.
Additional reporting by Mark Chediak.
Chart: Distribute the Wealth
Publicly launched in 1999, Cosm is a set of open protocols and applications designed to allow computers all over the world to work together on projects. Cosm is an example of a distributed system built on top of a traditional operating system. It scales from handling systems on a local network to systems built on the Internet.
- Beowolf Project
Beowolf is a combination of software and protocols designed to work with fine-grained and only with a very small, very fast computer network. It does not scale to thousands of machines working together or to slower communication mechanisms like Internet transmission.
- Globus Project
Globus is a research and development project focused on enabling the application of grid computing concepts to scientific and engineering computing. The grid is an emerging infrastructure protocol that enables the integrated use of remote high-end computers, databases, scientific instruments, networks, and other resources.
Condor is a software system that creates a high-throughput computing environment by harnessing the power of a cluster of networked Unix workstations. The key appeal of Condor, however, is its ability to effectively harness nondedicated, preexisting resources in a distributed ownership setting, such as idle computers in office labs.
© Mithral Inc.
All Rights Reserved.
Mithral® and Cosm® are trademarks of Mithral Inc.