October 28, 2009, 4:48 PM — In 2007, travel booking and expense firm Concur Technologies was growing at a rapid clip, but maintaining the IT infrastructure to serve its customers travel-booking and expense-tracking needs required an increasing amount of manpower and time.
The company regularly pushed out changes to its software, issuing a major feature release, which often encompassed more than a week of preparation, every month, and smaller bug fixes several times a week. The updates required patching the software on each of the physical servers in the company's data centers, an activity fraught with the potential for manual errors.
When Concur needed to expand its server capacity to match growing demand for its software-as-a-service (SaaS), the firm had to rush-order the computers, have them delivered to the proper data center, and configure them. The process generally took weeks, says Craig Baughn, vice president of Concur's hosting services.
"Three years ago, if I needed a server because of a capacity issue, we had to scramble around," he says. "There was a whole series of physical events that had to take place."
[For timely cloud computing news and expert analysis, see CIO.com's Cloud Computing Drilldown section. ]
That's a problem for a company that processes travel expense on behalf of 9,000 customers and 10 percent of the world's business travelers. So, beginning in 2008, Concur embarked on an internal program to embrace the core concepts of cloud computing for its data center. The company virtualized its data centers -- no longer running its core applications on physical servers but on virtual machines -- and focused on making its capacity elastic, expanding dynamically with the needs of its clients and internal groups.
"The owners (internal departments) of the products or the SaaS services don't always want to know the super technical nuts and bolts of the platform, but they do want to know that their requirements are delivered quickly, effectively, securely and with higher service levels," Baughn says.
Time, Cost and Compliance Benefits In early 2008, the company began implementing its private cloud, choosing VMWare for its virtualization solution and BMC's BladeLogic software tools to manage its virtualized data center. The result: A year ago, Concur had only 5 percent of its data center's servers virtualized; today, Concur has virtualized more than 42 percent of its servers. Baughn hopes to hit the 80-percent mark in two more years.
The improvement has been dramatic, Baughn says. Errors have almost been eliminated in deploying updates to applications, with incidents caused by configuration mistakes down 85 to 95 percent. And spinning up additional capacity is a much faster process. "We have gone from weeks to days," he says.