eWEEK content and product recommendations are editorially independent. We may make money when you click on links to our partners. Learn More.
With the holy grail of utility or grid computing as their quest, Oracle Corp., IBM and Veritas Software Corp. are each working on technologies tied to their respective platforms that automate load balancing management—a key milestone in the journey.
Oracle, of Redwood Shores, Calif., is beta testing Oracle Database 10g functionality that extends the capability of OEM (Oracle Enterprise Manager) to alert database administrators when workloads fall out of sync.
With the beta, which company officials call “Resonance,” OEM would alert the clustered database itself. The database would then rebalance itself to meet predefined service levels, said Benny Souder, vice president of distributed database development for Oracle.
Resonance, in beta testing for two months, is expected to be generally available later this year, possibly as early as next month.
Using the Intelligent Agents within OEM, the database would be able to respond to the code and automatically add server blades, for instance, to rebalance workloads in active or passive mode. Administrators, for their part, would be able to define priorities within the Resonance feature that will enable it to prioritize individual processes, Souder said.
“Grid is not some big database in the sky,” Souder said. “Grid is a smaller number of larger resource pools” from which users can get economies of scale, he said.
Similarly, IBMs Santa Teresa Labs has a project under way code-named SLEDrunner (Service Level Enforcement Discipline). Rather than load balance on the database, this technology would automatically load balance in a utility computing environment. IBM expects SLEDrunner technology to first work its way into the companys SAN Volume Controller, officials said. It could also be extended to databases.
In future versions of DB2, the Armonk, N.Y., company plans to capture workload and use its Query Optimizer engine to optimize the database for index recommendation, according to IBM Fellow Pat Selinger.
Although experts caution that vendors must take care not to overly enable active-session migration, the level of environment complexity is too great for DBAs to ignore, they say.
Veritas, meanwhile, is integrating storage load balancing and provisioning technologies it acquired into its Global Cluster Manager. In addition, the next version of Veritas Cluster Server, due next year, will raise the number of nodes supported from 32 to 250 and add workload management capabilities.
According to officials of the Mountain View, Calif., software provider, Veritas will soon announce a massive enhancement of its core storage management and virtualization technology that will allow customers to automate load balancing and policy-based data movement. The beefed-up offering will integrate with Veritas current availability, performance and automation products to help customers assess their storage management resources.
Despite the movement from vendors, some users are not ready to buy into grids as database operations—particularly those with a high degree of transactional activity.
“Our workload is fairly constant,” said Michael Heaney, database manager for The Institute for Genomic Research, in Rockville, Md. “Theres no equivalent of a Christmas rush in which traffic spikes so high that wed need to add some temporary computer power, which mitigates against the usefulness of utility computing for us. … Color me skeptical.”
