 ‘Rightsizing’ is an oft-used term, and its meaning is sometimes less than precise. But sometimes it fits. Take the case of Fair Isaac, the leading provider of analytics and decision management technology. The company specializes in building analytic products, like the industry-standard FICO score, as well as market-leading tools and applications for business rules management, fraud detection, optimization and other critical business functions for customers in the financial services, insurance, retail and health care sectors.
Fair Isaac has achieved its market leadership, in part, through strategic acquisitions. But over the years, along with intellectual assets, the company has also acquired computer infrastructures. Ultimately, Fair Isaac found itself running a number of different data centers and more than 5,000 servers distributed globally.
As Christopher Rence (pictured), Fair Isaac’s Chief Information Officer, explains, growth meant headaches. “Fair Isaac has been growing organically and by acquisition for about 12 years,” Rence explains. “That process meant that we had server rooms, data centers, network closets… a lot of different technologies spread around the world. That’s expensive but also hard to manage.”
With many of the data centers at or near capacity—and many of the servers at or near end of-life—Rence knew his team needed to refresh, standardize and consolidate infrastructure. Improving system manageability and operational efficiency (i.e., cost savings) were also considerations. And finally, as a 52-year-old pioneering company with its eyes on the future, Fair Isaac is focused increasingly on reducing its carbon footprint.
The task required the rationalization of 16 separate data centers that were at or near capacity and the replacement of an oversized fleet of servers, many near end-of-life, with more efficient systems that could deliver better performance in a smaller footprint.
After evaluating a number of vendors, Fair Isaac concluded the HP BladeSystem was the best fit for its consolidation plans. One big factor was manageability. Fair Isaac had already implemented the HP Systems Insight Manager framework, and the ability to use known, trusted tools played a big role in the decision to standardize on HP blades.
Another advantage was space utilization. Using blades enabled Rence and his colleagues to fit more servers into a new 4,000-square-foot facility than in one of their old 10,000 square-foot data centers.
When Fair Isaac decided to virtualize in order to optimize space savings further while gaining greater efficiency and flexibility, the company turned to HP Services for help. HP’s architectural consultants worked with Fair Isaac engineers to perform an infrastructure assessment and prepare an architectural design. For hardware, the team began with HP ProLiant BL45p G2 blade servers with 32 gigabytes of RAM. VMware was chosen for virtualization.
An HP Software specialist helped Fair Isaac configure management software, including HP Systems Insight Manager, HP ProLiant Essentials Management Software, HP ProLiant Rapid Essentials Deployment Pack, and HP Instant Support Enterprise Edition.
Next, the team tested the migration by transferring the contents of 40 older development servers to the new HP BladeSystem. After a “wildly successful” 30day pilot test, which was interruption-free for customers and had only minimal impact on internal users, Rence and his team settled down to the process of server conversion—and the shuttering of now-unneeded data centers.
The payoff
“We’ve gone from 5,000 devices to about 2,600,” Rence says. “We’ve leveraged the HP c-Class blade chassis to collapse the physical environment down to a virtualized one.” In one case, Fair Isaac was able to configure 800 virtual machines onto 45 physical servers.
Virtualization has also enhanced flexibility. “When we need to shut something down or bring something up, we can do that through virtualization,” Rence says. “For example, when a development group finishes we can back up the environment and shut it down and then reuse that resource for something else. It means we don’t have a lot of idle environments sitting around waiting for something.”
|
Printable version
