SAN FRANCISCO — Environment Canada is completing the test phase of a project to increase the scaleability and availability of the Canadian weather archive for research scientists and the general public.
The federal department hopes to bring a deployment of Oracle‘s Real Application Clusters into production in January of next year, according to Frank Lauritzen, manager of database administration at the National Archives and the Data Management Branch. Lauritzen is attending OracleWorld 2003 this week to keep abreast of the company’s latest products and services.
The Canadian National Climate Data and Information Archive has been running on an Oracle database since the mid-1990s, Lauritzen said, and is currently using 8i. Environment Canada will be launching the Real Application Cluster, which has a number of servers share the workload of a single application, using Linux on Dell hardware, he said.
Canada has been compiling its weather-related archives since the 1800s, Lauritzen said. Information now stored on Oracle was once entered on punch cards before moving to tape storage. The records include an array of information such as temperature, station pressure and water quantity data, among others.
The Oracle RAC project is expected to increase the reliability of the archives for both global weather researchers and everyday Canadians, who have had access to the archive through the Web since last year.
“”We will use every shred of compute power out there, because there is so much data to store,”” said Lauritzen, who estimated that 1TB of information has already been archived. “”It’s the logical place to go if you’re already on Oracle.””
Oracle is using this week’s conference to push the latest version of its database, 10g, which is designed to increase the way organizations can distribute the workload of major computing tasks. Andy Mendelsohn, vice-president of Oracle’s server technologies division, said the company has slowly realized its database needs to leverage the falling cost of storage. “”The way people run IT shops is backward,”” he said, referring to the dedicated servers that run HR, financials and other databases. “”How do you hard-wire all your I/O at the disk level? You need to balance that in order to avoid hotspots.””
Lauritzen said Environment Canada had already adopted the idea of a single database and shared resources as a best practice when Oracle launched 9i. He also said the archives don’t have the same kind of bottom line-driven availability issues that plague private sector firms.
Though Lauritzen said he doesn’t see Environment Canada creating a complete grid computing environment for the archive anytime soon, he admired the manageability features in 10g and the Application Server. There might eventually be the potential, he said, for the department to create a grid with other research groups.
“”You could have these kinds of service bureaus around the world, where if I have a computing problem I need resources for, you put that out for bid,”” he said. “”But that’s a long way off.””
Environment Canada is considering making more weather radar data available in the archive, but Lauritzen said this could add up to a terabyte a year, and the scientific community may need to look at 100TB before it could begin to isolate patterns and analyze them.
OracleWorld 2003 continues through Thursday.
Comment: [email protected]
