While Oracle prepares to build a larger, stronger business software vendor following its acquisition of PeopleSoft, the company is turning to industry partners to help build something else: standardized grids of computing resources.

At last week’s OpenWorld conference in San Francisco, Oracle

announced Project MegaGrid, an alliance with Dell, Intel and EMC to develop a common approach to organizing IT infrastructure. The partners have set up a prototype environment running Oracle’s grid computing database, 10g, on Intel-based Dell servers with EMC storage products at a data centre in Austin. The project will involve configuring and testing a blueprint for grid computing systems that will be available in the form of white papers on each vendor’s Web site.

Oracle has been pushing grid computing, a way of sharing large computing workloads among a string of connected servers, since the debut of 10g at OpenWorld 2003. Its product strategy is based on the premise that customers will integrate various pieces of software — either from its E-Business Suite, SAP’s or PeopleSoft’s applications — and tie them to its database. The demand for better performance from integrated applications would create a need for grid computing, according to Oracle executives.

All four partners are also members of the Enterprise Grid Alliance, a consortium that works on technical standards, but the idea behind MegaGrid is to get grids into enterprise shops, said Sohan Demel, director of server technologies at Oracle Corp. “”We want to lead customers on this journey,”” he said. “”We wanted to validate how big iron can be done on a grid infrastructure.””

Dell is contributing dual Intel Xeon and four-way Intel Itanium processor-based PowerEdge servers to the project, in part because the company sees Oracle’s grid strategy as synonymous with Dell’s own scalable enterprise vision, said Judy Chavis, Dell’s director of business development. Project MegaGrid is a good way to show how Itanium servers, for example, can run one million transactions per hour, she said.

“”It’s not academia, it’s not high-performance computing, it’s real-world,”” she said. “”Obviously, when you talk about these applications, they’re going to be running for a long time.””

While it may be possible to create similar environments running AMD processors and server products from Sun Microsystems, Project MegaGrid could simplify the process for many enterprises, said Neil Blecherman, Intel Corp.‘s director of corporate development.

“”Other companies are free to do — on their own nickel, and their own time — what it took us a year to do,”” he said.

Because the prototype grid is running on open source, the choice of partners was deliberate, Demel said. “”When you talk about Linux, Intel and Linux-based servers just resonate with customers in terms of low-cost deployments,”” he said.

Shawn Douglas, sales director at EMC, said the white papers are useful because they show customers how grids could be constructed based on either storage or server needs. EMC’s contribution includes its Clariion CX and Symmetrix DMX networked storage systems. “”You don’t move customers to this overnight,”” he said, adding that the best practices developed in Project MegaGrid will reduce the time to implement. “”Grids aren’t bought, they’re built.””

Oracle promoted Project MegaGrid with several sessions at OpenWorld, and will follow up by discussing the best practices at an “”Architecture of the Future”” road show across North America. The next phase of the project will have partners come up with ways to add multiple types of workloads concurrently, Demel added.

Comment: info@itbusiness.ca

Share on LinkedIn Share with Google+