Distributed & Grid Computing

You probably know that your desktop computer (like most others) is only utilized about 5% the CPU stays idle most of the time. Distributed computing uses this idle time and links many machines together to perform mammoth tasks that previously only super-computers could do.

 

Peer-to-peer networks

 

The traditional client-server Internet model ("client" asks for and receives information from "server") is beginning to give some ground to peer-to-peer (P2P) networking – where all network participants are approximately equal. The primary advantage of P2P networks is that large numbers of people share the burden of providing computing resources (processor time and disk space), administration effort, creativity and legal liability. It's relatively easy to be anonymous in such an environment and it's harder for opponents of a P2P service to bring it down.

 

Napster was one of the best-known examples to date, though it's not a completely P2P system because users register their file libraries with a central server when they log on to the service. Because of this central link, it was possible for opponents to force Napster to stop facilitating free music downloads. This was done by legal action on behalf of the music industry, to protect copyright infringement. However, true P2P networks such as Kaaza and Morpheus now operate with impunity, because there is no central server, They are now generating traffic many times the rate of Napster at its peak.

 

P2P has some limitations

 

P2P does have problems. The primary disadvantage is the tendency of computers at the edge of the network to fade in and out of availability. Also, accountability for the actions of network participants could be a difficult problem. Several high-profile implementations have shown that architecture, security, and systems management issues are difficult to control. For these reasons, system managers often prefer to operate P2P systems as separate isolated entities. But, doing so is often impossible for practical applications.

 

The increasing reach of P2P beyond the desktop and into distributed computing fringe may further magnify these stumbling blocks. But that will not hold back the rising tide of P2P implementations. Dramatic cost savings and the ability to do parallel processing will continue to drive P2P use higher.

 

Distributed Computing

 

These days, there are more tools than ever to help companies harness unused computing power in the hundreds of PC being used by their employees. The enormous power of P2P computing is increasingly being blended with its more complex cousin, distributed (or grid) computing.

 

Traditionally, there have been three categories of "distributed" computing:

 

  • Cluster computing: Similar machines (generally servers of similar power and configuration) are joined to form a virtual machine. Linux clusters are good examples.

  • Peer-to-peer: Many desktop computers are linked to aggregate processing power. The distinguishing characteristic is the machine itself, which almost exclusively is a low-power client PC. Often, the link is via the Internet.

  • Distributed computing: Increasingly known as grid computing, this approach connects a wide variety of computer types and computing resources, such as storage area networks, to create vast "virtual" reservoirs of computers serving geographically widely separated users.

 

Although clustering retains a distinct niche, the line between P2P and distributed computing is blurring. One reason is that many distributed computing software vendors are making it easier to incorporate a mix of PCs, Macs, Linux and Unix servers, and even high-end multiprocessor servers as nodes in a peer-to-peer computing system.

 

Grid Computing

 

Grid Computing can be defined as applying resources from many computers in a network to a single problem, usually one that requires a large number of processing cycles or access to large amounts of data.

 

The computational power grid is analogous to electric power grid. It allows the coupling of geographically distributed resources to offer consistent and inexpensive access to resources irrespective of their physical location or access point. The Internet or dedicated networks can be used to interconnect a wide variety of distributed computational resources (such as supercomputers, computer clusters, storage systems, data sources) and present them as a single, unified resource.

 

At its core, Grid Computing enables devices—regardless of their operating characteristics—to be virtually shared, managed and accessed across an enterprise, industry or workgroup. This virtualization of resources places all of the necessary access, data and processing power at the fingertips of those who need to rapidly solve complex business problems, conduct compute-intensive research and data analysis, and operate in real-time.

 

A company with slightly fewer than 2,000 desktop computers can harvest nearly 1 teraflop (one trillion floating-point operations per second) of computing capacity. Even better, the company can capture that power from computers it already owns that sit idle at night and work at less than full capacity during the day.

 

Universities and research institutions have long used grid-computing technology, but recently it is also making fast inroads into the business market. IBM has recently come up with operating standards that will generate many new business applications and will cause wide proliferation.

 

Internet Grid Computing

 

The Internet is evolving beyond e-mail, content, and electronic commerce. It is becoming a true platform, combining the qualities of service of enterprise computing with the ability to share distributed resources across the web — applications, data, storage, servers, and everything in-between.

 

  • Enabling devices: The Internet extended the range of things that personal computers can do, taking advantage of its communications functions to put servers to work storing backup files, serving personal web pages and blocking spam. Still, the personal computer, personal digital assistant (PDA) or other connected devices do not have transparent access to all types of programs on all kinds of platforms. The applications they access are not necessarily integrated. Grid computing holds the promise of bringing this sort of power and capability to Internet-enabled devices. In a wireless world, this grid could enable even simple devices – such as pagers – to access the power of computers across the network in a meaningful way.

 

  • Intense collaboration:  Today, online collaboration provides document sharing and allows discussions. This process will change with grid computing, which will facilitate joint use of complete applications. Though this has tremendous use for sophisticated collaboration applications, even mundane applications will benefit. For example: patterns can be detected in the use of financial, home improvement, event planning and other applications, and all connected applications will benefit from the options identified, and choices made, by people with similar needs. Many sorts of simulation, including prototyping, can be made possible by grid computing. This will encourage new relationships and new communities.

 

In addition to new capabilities, there is great potential for cost savings. Using spare cycles (idle time on your computer), paralleling attacks on problems, accessing via simpler devices and using only the application functions you need should make most information technology applications more economical and affordable in the future.

 

Applications

 

The whole area of distributed computing is a hot bed of significant development that is expected to generate amazing advances in the next few years. And the first big applications are already here.

 

United Technologies, the $28b manufacturing conglomerate, is equipping more than 100,000 Wintel computers with its own peer-to-peer software to do scientific calculations and solves complex modeling problems during off hours. The project is an expansion of what the company's Pratt & Whitney aircraft engine division has done to phase out a Cray supercomputer with 5,000 Sun Unix workstations, to perform design simulations for aircraft parts. The result: 85 percent utilization for each workstation.

 

With 20,000 PCs in Pratt & Whitney, the P2P project is expected to cut in half the time and money it takes to develop turbine engines and other aircraft parts, mainly by eliminating multimillion-dollar prototypes. Previously, it could take $1 billion and five years between the time an engine is developed and certified.

 

Universities and research organizations have used similar peer-processing approaches to solve complex scientific problems. The most famous is a project called [email protected], which uses the computers of volunteers across the globe to search for life on other planets.

 

An Intel philanthropic P2P Program helps to combat life-threatening illnesses by linking millions of PCs into what is predicted to be the largest and fastest computing resource in history. This "virtual supercomputer" uses P2P technology to make unprecedented amounts of processing power available to medical researchers to accelerate the development of improved treatments and drugs that could potentially cure diseases.

 

If you feel P2P and grid-computing are important, you might enjoy reading a recent book: "Peer-to-Peer: Harnessing the Power of Disruptive Technologies", a collection of chapters written by the people who are driving the state of the art in the P2P space. This book is NOT about how to build the next Morpheus or Kaaza system. Rather, it aims to get its readers thinking about what happens when information systems shift away from client-servers towards the peer-to-peer model.

 

Related links:

 

PCs as supercomputers:

http://www.internetweek.com/newslead01/lead052901.htm

 

Intel philanthropic P2P website:

http://www.intel.com/cure/

  

The peer-to-peer working group

http://www.peer-to-peerwg.org/

 

Book: P2P - Harnessing the Power of Disruptive Technologies.

http://www.jimpinto.com/reading.html#P2P

 

Grid computing info center:

http://grid.ihep.su/indexGrC.html

 

Jim Pinto is an industry analyst and commentator, writer, technology entrepreneur, investor and futurist. You can email him at [email protected]. Or look at his poems, prognostications and predictions on his website: www.JimPinto.com