Data Center Consolidation Sparks Converged Infrastructure Deployment

SIRVA, an international relocation and moving solutions provider, determined that acquisitions are a good way to grow a business. However, one repercussion from the strategy was inefficiency: Duplicate business functions were performed in various offices scattered across the United States. In streamlining its IT infrastructure, the corporation found itself at the forefront of the movement to consolidate data center products, a position that, to date, has paid dividends for the corporation.

SIRVA supplies relocation and moving solutions to consumers, corporations and governments. The company, which has 2,600 employees, offers a variety of relocation services, including program development and management; home purchase and home sale; moving; and mortgage services. Allied, DJK Residential, Global, North American International, SIRVA Mortgage and SIRVA Settlement are a few of its brands. Each day, the corporation, whose annual revenue is $2.3 billion, helps more than 1,000 families in more than 175 countries relocate.

After a series of acquisitions, the business found itself in 2007 with four data centers, located in Chicago, Cleveland, Fort Wayne and Minneapolis. The centers had several hundred terabytes of storage and a hodgepodge of computer systems: Unix servers, Wintel-based systems and even an IBM mainframe. After a thorough evaluation, the services company decided to consolidate its computing infrastructure into a primary production data center in Fort Wayne and a disaster recovery, testing and development site in Cleveland. "We thought we could use new technologies, such as virtualization and a converged infrastructure, to reduce our data center operating costs as well as provide us with more visibility into how our different systems were functioning," says Chuck Schmayel, VP of security at SIRVA.

The relocation company then went out to the marketplace to determine what vendors had to offer. After examining products from Dell, HP and IBM, SIRVA opted to go with HP. "We already had been using a fair amount of HP equipment," says Schmayel. The first phase in the relocation company's grand plan was to install HP ProLiant BL490c and 460c G6 server blades in HP BladeSystem c7000 enclosures, along with VMware's virtualization software.

SIRVA started off using storage solutions from Network Appliance but eventually switched to HP StorageWorks EVA8400 and 4400 Enterprise Virtual Arrays. "The cost of the Network Appliance products was high, and they did not offer us as much systems performance visibility as we wanted," notes Schmayel.The HP StorageWorks EVA4400 was deployed as a network attached storage system for the mainframe and also acts as a virtual tape library. The latter meant disk-to-disk backups, which led to a 50 percent reduction in tape costs. Another plus from the move to the virtual tape system was reduced retrieval time. Rather than a couple of days, the time to restore information was cut to a few hours.

In 2010, the multimillion dollar project hit some major milestones. While the initial plan was to complete the process in 18 months, much of the work involved with building the new data center was finished after about eight months. Because of the change, the company was able to consolidate 600 servers to 130 servers and realize a10-fold reduction in server racks, from 30 to three. Another plus was data center productivity gains. Using virtual machines cut the time needed to provision a new server from four weeks to four hours.

The end result was significant reduced operating costs. With fewer physical servers, maintenance and licensing costs went down about 30 percent, and power and cooling costs dropped 40 percent. In addition, the company cut its staff by 30 percent--largely individuals who did not want to relocate.

Also, the IT infrastructure became more resilient. The new system allows SIRVA to automatically load-balance virtual machine resources across blade chassis, so, for instance, they can automatically move workloads from failing hardware to other functioning devices.

The company did encounter a few hiccups."Because the infrastructure is so different, we had to invest a lot of time training of the staff," explains Schmayel. In some cases, key technicians attended training classes that lasted a full week, which stretched the staff thin.The company is about 85 percent finished in moving all of its applications and expects to complete the process later this year. "We encountered some challenges along the way but found our effort to be worth it," concludes Schmayel. "Our IT infrastructure is more efficient, resilient, dynamic and cost-effective than it was when we started the process."

You might also like Informed CIO: Data Center Consolidation this which is available with a subscription to Network Computing Pro.