The Hindu Business Line : Computing power on demand

SUDARSAN, a sales representative in a small consumer durables firm, is a frustrated person. Reason: Too often, the data he seeks does not appear on his computer. He is not alone. Officials in most companies that have tonnes of data sitting in various servers at different locations face a similar problem. However, leading infotech companies say they have a solution to this hitch.

Welcome to the age of grids, or grid computing. Simply put, grid computing ties together processing power and storage in a large cluster of technology resources, which can be allocated to applications on demand. The benefits include efficient use of processing power, since no computer would stay idle whenever there is work to be done, says Brajesh Goyal, Principal Product Manager, Grid Computing, Oracle Corporation.

"After the Internet, grids may be the next big thing in high-performance computing," he adds.

Global companies, including Oracle, IBM and Sun Microsystems, have long been trying to combine hundreds of computers seamlessly within a company, to make processing power available on demand. Not very different from electrical power.

As in peer-to-peer computing, grid computing allows users to share files. However, unlike peer-to-peer, grid computing also allows many-to-many sharing of files and other resources, as well.

Grid computing connects pools of computers, storage and networks, enabling public and private sector companies to allocate and adapt resources based on changing business needs.

Though the concept started about five years ago, grid computing is gaining momentum only in recent months. It will take a few years for companies to realise the actual benefits of grid computing, says Goyal.

Data on demand

Grid computing works like a utility function. Just as electricity is made available whenever there is demand, data should also be available whenever it is required. Grid computing enables this, he says.

In other words, it does not matter where the data resides in a company or which computer processes one's request. Users would be able to access unlimited information or computation whenever they want, says Goyal.

Different firms use different terms to describe grid computing. Some call it group computing, distributed computing, large-scale cluster computing and even "a form of peer-to-peer computing". However, it all boils down to the same - on demand computing. As Goyal explains, a grid user essentially sees a single, large virtual computer.

As for the opportunities afforded by grid computing, Goyal says that it is as big as the IT industry itself. Any company that deploys IT would be forced to adopt grid computing sooner or later, he says.

At present, most applications draw on resources from a single server. When business processes change, companies will need to buy new server capacity, write new integration software and undertake further testing. This infrastructure may take a long time to change, to reflect new business requirements.

However, grid computing introduces workload management capabilities that enable applications to share resources across several servers.

Data processing capacity can be added or removed on demand. By centralising resources and treating them as a single unit, grid computing helps create a far more secure infrastructure than any other architecture, says Goyal.

Drawing comparisons, Goyal says that while other vendor grids require custom applications and new infrastructure, the Oracle Database 10g and Oracle Application Server 10g can pool servers and storage into a flexible, on-demand computing resource. All systems, including mainframes, Unix, Windows and Linux servers, can be used to build an enterprise grid using the Oracle 10g software.

An Oracle business white paper says that existing enterprise applications such as enterprise resource planning (ERP), customer relationship management (CRM) and supply chain management (SCM) benefit from the higher utilisation, reliability and security offered by Oracle grid computing.

For example, towards the end of the financial year, there is greater demand for order processing, while business intelligence systems idle.

And at the beginning of the following quarter, business intelligence applications require more resources while order processing drops. The grid allocates servers accordingly.

Gradual migration best

Goyal advises companies to migrate to grid computing in a gradual manner. They could start by running a small cluster of servers and storage on one application. As the company decommissions old systems, it could add new capacity to the grid before moving applications over. Linux provides a cost-effective operating system that can be used throughout the grid, says Goyal.

The second step is to consolidate infrastructure in one or a few data centres. With this, the company will have fewer grids and larger pools of resources.

The next step is to automate the grid so that the company can manage it effectively as it grows. As enterprise grids can have potentially thousands of servers, a grid is simply too large to manage server-by-server. Oracle has automated the day-to-day maintenance and provides for a centralised management console. This means the entire infrastructure can be managed as one large computing system. One or a few administrators can maintain even the largest grid data centre.

The final step in grid formation is to adapt an application to use the grid services architecture, he says.

According to Goyal, Oracle has been using grid computing for the last few years to improve its business processes. For instance, Oracle's Application Demo Systems (ADS) is used by its global sales organisation. ADS runs the entire e-business suite - some 180 modules - in a single instance, and provides 450 copies of the environment for the sales force!

Global alliance

Global IT players recently joined hands to promote grid computing in large organisations. It is called the Enterprise Grid Alliance (EGA) and its founding members include Oracle, Hewlett-Packard, Intel, NEC, Sun Microsystems and Fujitsu Siemens Computers.

The EGA is an open, non-profit, vendor-neutral organisation formed to develop enterprise grid solutions and accelerate the deployment of grid computing in enterprises.

The alliance fosters open solutions and best practices among companies using grid computing.

Picture by S. Thanthoni

raja@thehindu.co.in

Article E-Mail :: Comment :: Syndication :: Printer Friendly Page

Stories in this Section
Slowing down

Computing power on demand

Some pencil truths

At the stroke of a mouse

Here's the key

Trouble with macros