Columns

Online, docs take care

• By Kathleen Ohlson
• July 1, 2005

Doctors at the CareGroup Healthcare System were looking for a better way to access patient information, so the Boston-area health service developed CareWeb, an intranet system that gives physicians real-time access to patient records. The site includes a portal for patients to communicate with their doctors and access personal medical records. CareWeb’s infrastructure is based on Windows 2000, Windows 2003 and Unix servers, Exchange 2003 mail servers, Internet Information Server Web servers, and SQL Server 2000 and Oracle 9i database servers for storing hundreds of gigabytes of critical patient data. CareWeb gives users access to 300 applications and can handle more than 240 concurrent users.

“It’s not easy to find what the problem is and where it’s going. There’re so many tools and scripts, and it takes a long time to collect data and find out what’s causing bottleneck,” says Ayad Shammout, CareGroup’s senior database administrator and analyst. It would take days or weeks to investigate the cause of a performance bottleneck using SQL Server’s built-in performance tools. “You have to troubleshoot and go step by step. It takes time, and in our environment, that’s not acceptable.”

Curing a performance headache
CareGroup reviewed vendors for an app performance management tool, selecting Veritas i3 for SQL Server. Although other vendors’ tools featured real-time monitoring and real-time alerts, they lacked the ability to save historical data, requiring different tools to fill the gaps, according to Shammout. He particularly wanted a tool with “minimal or zero overhead” and one that would require very little configuration and customization.

According to Shammout, Veritas i3 started collecting data immediately, and within a few minutes, the tool showed the overall health of CareGroup’s servers. The tool featured colorful graphics and showed the colors’ meanings, as well as how servers performed in the past and will perform in the future. “I can focus on one performance issue, drill down and find the inquiry,” he says.

More important, Veritas i3 finds problems with almost no downtime, reaching its goal of 99.99- percent availability. “It’s saved me a lot of time,” he says. “I’m not spending an hour or two hours on tuning and monitoring; I’m relying 100 percent on that product.” For example, a problem that was first thought to be related to SQL Server turned out to be caused by CareGroup’s billing information database. Veritas i3 identified the problem with disk layout on the back end. Shammout discerned that only two of 18 disks were active and reconfigured the disks to improve application performance.

Veritas i3 automatically sends alerts via e-mail or pager when a problem is detected. “When I get an e-mail from Veritas, something really happened. There aren’t any false alerts,” Shammout says. “It’s helped a lot; rather than spending hours and days, it takes me minutes.”

Rolling in production
CareGroup initially installed i3 6.5, and participated in the beta program for 7.0, rolling out the new version in its production environment. The i3 7.0 now provides real-time alerts and features a Web interface so it can be accessed from any machine through a Web browser. In addition, systems don’t need to be rebooted and have scheduled downtimes for IT to install the software, apply a patch or perform an upgrade.

Implementing Veritas i3 for SQL Server eliminated the need for different departments to meet to work out one problem. “Beforehand, we’d pull the server, storage and DBA teams together to sit down to find the problem,” he says. According to Shammout, other CareGroup departments are investigating implementing Veritas i3 to solve similar performance issues.