New approach is the first to optimize an entire database workload in one go.
JERUSALEM, Aug. 10, 2010 -- DBSophic, an innovative provider of performance management products for database-centric applications, announced a new approach to database optimization, known as "workload tuning". Unlike traditional "query-by-query" tuning techniques, workload tuning optimizes the entire database workload automatically. Since a workload may include millions of queries, the process runs offline, incurring no load on the production environment. With support for applications running on SQL Server, the new approach is implemented in the company's flagship product, Qure™.
Qure has been successfully deployed at Clalit Health Services, one of the world's largest HMOs, serving over 3.9 million people. At Clalit, delivery of state-of-the-art medical care is facilitated by a large and complex IT infrastructure that includes over 500 SQL Server databases. Key SQL Server applications include administrative systems and the patient-information system in Clalit's hospitals.
"Our IT applications serve a large base of doctors, healthcare workers, and of course patients," explains Doron Ytzhaki, CTO of Clalit Health Services. "Providing our users with the fastest application performance is a top priority for us. To help achieve this, we're using Qure throughout our SQL Server environment. Thanks to Qure, we've already achieved major performance gains on both the administrative systems in our data center and the patient-information system in our hospitals."
At Clalit, dramatic improvements in application performance have also translated to a significant IT cost savings, as Ytzhaki explains: "Like other large sites, we have consolidated our many SQL Server databases onto a centralized server farm, to ease manageability and control costs. By helping us get the most from our existing hardware, Qure has saved us server hardware upgrades, DBA resources, and improved the performance of the DBA group."
Ami Levin, DBSophic's CTO, explains how workload tuning and the company's Qure solution work: "To provide automated optimization of millions of queries, thousands of stored procedures and other database objects, Qure runs against a copy of the production database. The workload to be tuned is represented by a standard SQL trace of production activity.
"After Qure's analysis process generates the recommendations," continues Levin, "Qure benchmarks their actual performance improvement against the copy-of-production. Qure does this by replaying the workload, applying the recommendations, and then replaying the workload again to measure the actual performance gain (in Duration, I/O, and CPU).
"Finally, Qure provides the DBA with executable scripts for applying the recommendations to production. Recommendations may be applied by the DBA selectively or in bulk and are always accompanied by detailed explanations and undo scripts."