Ellison on Grid Computing: It's Invincible, It's Inevitable

ORACLEWORLD, SAN FRANCISCO -- Sometimes you have to dial back the clock to understand today's breakthroughs. Speaking to thousands of attendees at Oracle's customer and partner conference, Chairman and CEO Larry Ellison launched Oracle's Grid Computing initiative, but to explain its significance, he dialed back the clock to 1964, to the advent of mainframe computing. "The industry has been on a quest to build bigger and bigger mainframes for the last 40 years. We've been chasing the same dream of building the fastest computer in the world," he explained. "After 40 years, now there's an alternative to the one, big server approach. It's enterprise grid computing." Ellison expounded on the "one, big server approach" and the issues that have plagued it: limited capacity, high cost, and limited reliability. When the one server goes down, the application also goes down. The answer, he advocated, is moving to grid computing, a new architecture that connects low-cost computers, storage and networks together to act as one computer, but at a fraction of the cost and with ultimate reliability -- there is no single point of failure. While grid computing has been used in scientific research, it has never been applied to business software, which is Oracle's focus as it rolls out its Oracle 10g Grid Computing software starting this week. "The beauty of Oracle's enterprise grid computing is that our software creates the illusion of one big computer." He went on to explain the immense cost savings and the significance of Oracle's ability to run packaged business applications, such as SAP, unchanged, which means that any business can take advantage of the new software without changing code. "It's capacity on demand. Plug another server into the grid and the application runs faster and more reliably, and the capacity is inexpensive." To underscore his point, Ellison said that comparing a single IBM mainframe with an Oracle grid built with inexpensive Intel-based blades will result in a 30-to-1 cost savings for customers. As Ellison reiterated today, the business efficiencies are the primary reasons to make the move to enterprise grid computing, but, "you have to be willing to pay less." Ellison referenced customers such as Electronic Arts and CERN in addition to Oracle, whose systems are easily handling escalating demands, but at a much lower cost due to both hardware and labor savings. For example, in moving to an enterprise grid, Oracle University was able to reduce 216 hours per week of work by its system administrators and DBAs to just two hours per week. Calling IBM's Computing On-Demand strategy a "financing option," Ellison emphasized that what Oracle will be shipping by the end of the year -- Oracle Database 10g, Oracle Application Server 10g and Grid Control -- are major technology advances. "We've been working on this for 12 years, starting with Oracle Parallel Server, then Real Application Clusters and now Grid. It's technology that keeps getting better and better."