Saturday, 21 May 2011

Grid Computers

Grid Computers


We all know a PC, probably heard of supercomputers. But something like a GRID computer is only known to a small section of ICT professionals. Little you hear about such machines via newspapers, or some via the internet when you look for it.

A grid computer is nothing else than a multiple number of the same class of computers clustered together. Often the connecting technology is called clustering. Internal a grid computer is connected through a superfast network and shares other devices such as diskdrives, printers, mass storage and mass memory. A sophisticated operating system takes care for the (load) sharing in computing and processing. (2)

Machines in a grid configuration cost a fraction of what a super cost: less than 10%.

If you use micro computers to combine into a grid. You will have a little less power but by adding more members you can generate computing power way beyond the limits of the individual machines.

The big difference with mainframes and supercomputers is that a grid or cluster computer is mostly build from a large number of self-contained computers. At least the separate machines could work stand alone. And the other big difference is: it can be build from low end computers like Intel pentium machines. And even Commodore 64's if you could change the OS to work parallel and amass enough C64's (2)

If you combine machines like IBM RN6000's into a grid or cluster you will have gigantic super power. An example of this is Blue Gene that beat the world champion in chess: Casparow in 1998

In fact a grid can easily grow into a super that is mostly a combination of parallel computers or a massive amount of cpu's mounted on separate cards, like the Cray machines

Look at a grid computer as if a million people can calculate faster together then doing the calculations all on them individually.

"With a million people you can create a road in one day, one worker needs a million days to do the same."

The image below is a very, very simplified example of this principle.

Each yellow dot represents a computer. Each computer has one single task: add two numbers.

The trick is while adding two numbers and passing that to the next row ( 2 + 2 ) the first row can do a new calculation again while the other is busy.
And as you see the final answer does not have to be computed by one single computer. In principle this is how supers and all other parallel computers work too! You will also see that this type of computing is useless when you only have to do some simple adding.

Another example that illustrates the sheer strength of a grid computer is when you draw a fractal (ref: mandelbrot) image. Each turn/part can be calculated separately because the high symmetry of the image:

As you might have guessed this class of computers has brethren in all computer stratae:

• Quantum computers
• Grid computers
• Supercomputers
• Mainframes
• Mini computers
• Microcomputers
• Terminals
• Embedded computers

The ranking of a grid computer could as you will understand be placed in all classes.

Grid computing enables the virtualization of distributed computing and data resources such as processing, network bandwidth and storage capacity to create a single system image, granting users and applications seamless access to vast IT capabilities. Just as an Internet user views a unified instance of content via the Web, a grid user essentially sees a single, large virtual computer.

At its core, grid computing is based on an open set of standards and protocols — e.g., Open Grid Services Architecture (OGSA) — that enable communication across heterogeneous, geographically dispersed environments. With grid computing, organizations can optimize computing and data resources, pool them for large capacity workloads, share them across networks and enable collaboration.

Evolution, not revolution

In fact, grid can be seen as the latest and most complete evolution of more familiar developments — such as distributed computing, the Web, peer-to-peer computing and virtualization technologies.

· Like the Web, grid computing keeps complexity hidden: multiple users enjoy a single, unified experience.

· Unlike the Web, which mainly enables communication, grid computing enables full collaboration toward common business goals.

· Like peer-to-peer, grid computing allows users to share files.

· Unlike peer-to-peer, grid computing allows many-to-many sharing — not only files but other resources as well.

· Like clusters and distributed computing, grids bring computing resources together.

· Unlike clusters and distributed computing, which need physical proximity and operating homogeneity, grids can be geographically distributed and heterogeneous.

· Like virtualization technologies, grid computing enables the virtualization of IT resources.

· Unlike virtualization technologies, which virtualize a single system, grid computing enables the virtualization of vast and disparate IT resources.

And because grid computing dovetails naturally with powerful developments in IBM like autonomic computing and e-business on demand, grid offers businesses a host of practical benefits for coping with — and taking advantage of — an on demand world.

No comments:

Post a Comment