An enterprise grid is a form of datacentre computing where low-cost servers are clustered to provide a scalable server farm. This allows additional processing power to be added or taken away by the addition or removal of servers. The key to its success is that existing applications should be able to take advantage of the grid without having to be redeveloped.
Research by Oracle and analyst firm Quocirca, released last week at Oracle Openworld in London, predicted that adoption of enterprise grids was likely to follow a similar pattern to the growth of the internet in the 1990s.
To date, computational power has been governed by Moore's Law, which states that processing power doubles every 18 months. It is this law that has driven the development of increasingly powerful datacentre servers.
However, Oracle president Charles Phillips, speaking at Oracle Openworld, said, "For the first time in 30 years, Moore's Law is reaching its limits."
Phillips suggested that with chipmakers such as Intel introducing multi-core processors to increase power, Moore's Law no longer applies. "This is an inflexion point in hardware design," he said.
Phillips believes the only answer is for users to develop enterprise grids to cope with the increasing demand for com-putational power, such as handling the monthly peak during a payroll.
However, enterprise grids may not be the only way to cope with users' needs for more processing power, said Rob Hailstone, research director at analyst firm IDC. "[Enterprise] grids will exist, but they may be too complex for some situations," he said.
Hailstone suggested that a less complex approach to datacentre computing would be a strategy based on a service-oriented architecture.
In this model, an IT infrastructure is built on many simple applications running on their own servers, each with their own database, that are able to share data with each other.
"Users will not need complex datacentres with service-oriented architectures," said Hailstone.