Grid is key to new IT economics


29 Jan 2004

One of the key challenges facing organisations today is the high cost of maintaining IT infrastructure. This can be linked to structural inefficiencies in most IT departments, mainly because of the need for businesses to buy far more capacity than they would routinely use, in order to be able to cater for predicted peak demand and possible system failure.

The inevitable consequence of this is a lot of computing power remaining idle for most of the time. This, in turn, leads to system complexity with a requirement for specialised tools and a high level of human intervention.

What is required is a new approach to infrastructure where all of an organisation’s computing power is available when and how it needs it, hence the advent of grid computing.

Simply speaking, grid computing can be defined as many small servers acting as one large computer. It has been an important trend in academic circles for some time, as they looked for ways to process large amounts of information. These academic grids tend to be used as number crunchers whereby they process large amounts of static information, be it simulations or specific research data. They also tend to use large computers such as mainframes.

This scenario, however, is not suited to a business environment where the emphasis is on real applications processing real data in real-time, which is how Oracle’s vision of enterprise grid computing came about.

With the new 10g database and application server, Oracle has delivered on this vision by creating the first infrastructure software designed for grid computing. This revolutionises IT economics by fully using all available computing power to provide improved levels of security, availability and performance on low-cost modular hardware.

The time is right to deliver enterprise grid computing because several trends within the IT industry are converging to make it possible. Three trends are:

Low-cost computing: There is a whole range of technology developments that are helping to bring down the cost of IT infrastructure. These include the growing use of increasingly powerful Intel chips and blades, improvements in high-performance network connectivity and developments in areas such as network storage. The development of Linux into a mainstream, low-cost, secure operating system that works with this new type of modular, low-cost, commodity hardware has been a powerful driver that has helped create an attractive basic building block for the enterprise grid.

Consolidation and virtualisation: This is at the heart of the grid. Consolidation is where all servers in an organisation are brought into the data centre to be clustered together so all applications are run centrally with a pool of resources that can be shared and managed to suit business requirements. Virtualisation makes this collection of small systems look like one big one. This is important to reassure users, but it allows standard off-the-shelf applications to be run without any changes.

Automation: There have been increasing developments in the area of automated management. This is crucial, because the key to a successful grid is the ability for resources to be shared and distributed seamlessly according to pre-defined business rules and demand. Even the most skilled systems managers would have difficulties managing this to the required service levels, without sophisticated automated tools.

Enterprise grid computing architecture does not require a large, upfront investment. Organisations can choose to move at a pace to suit them and because the grid does not require all servers to be the same, existing servers can be used as required. Should extra power be needed, inexpensive commodity components can be used.

Grid computing brings many benefits to an organisation. By providing a common pool of resources that can be changed on demand, a grid infrastructure provides the full flexibility to meet continually changing business needs. The days of business growth being constrained by limitations in IT infrastructure are now a thing of the past.

By pulling together all the components of an IT infrastructure into a centralised pool of resources, a grid provides benefits that were not available when these were managed and used independently. All of these resources can now be used at all times and diverted to areas of the business when the business dictates without compromising the improved levels of security a grid can offer.

By using all existing parts of the IT infrastructure, an enterprise grid fully employs those components that were previously idle as backup systems and actually improves the fault tolerance available by offering automatic switching of resources should any of the components fail.

The advent of enterprise grid computing heralds a new era in IT economics. For the first time, it is possible for companies to protect existing infrastructure investment, provide a better service, improve staff productivity and organisational efficiency, reduce downtime and cut costs.

By Andy Ellwood