Basel, Switzerland-based Panalpina Group, a logistics company, saved $1 million last year by consolidating seven outdated data centers into just one, but coordinating multiple data center migrations across branch offices throughout North America wasn't easy.
Data Center Manager of the Year finalist Armin Heinlein, the corporate vice president and head of IT competence at Panalpina, needed to centralize IT operations without disrupting services or bogging down end users with latency problems.
The data center consolidation project began in 2005, when many of Panalpina's smaller data centers needed infrastructure upgrades, new air conditioning, cabling and security systems. Upgrading the Toronto data center alone would have cost $500,000.
"We calculated that we had to spend a significant amount of money improving the infrastructure across all these sites," Heinlein said. "We were growing, and we needed additional horsepower. Consolidation made sense at that point."
Centralizing IT operations without latency issues
As a logistics company, Panalpina buys space from air and ocean carriers and fills that space with its customers. The company also tracks documentation for customers, prepares data necessary for U.S. customs, and gets confirmation when a freight has been released.
Prior to the consolidation, Panalpina had small data centers and IT closets sprinkled throughout North America. Although it was possible to move servers and infrastructure centrally, Heinlein's users would have experienced poor application performance by not having local servers.
The company relies on remote file servers and Citrix Systems Inc.'s technologies to deliver applications to a large number of mobile users, so poor application performance wasn't an option. Luckily, Heinlein identified latency the issues up front. "I was an end user working off a file server in New Jersey, and when I moved from New Jersey to Miami, I realized it was a problem," he said. "The network guys told me, but you just don't want to believe it."
Heinlein investigated wide area network (WAN) optimization products to boost application performance locally and, after evaluating vendors and solutions, selected Cisco Wide Area Application Services (WAAS), a SearchNetworking.com networking product of the year. Increasingly, data center managers who cope with bandwidth-intensive applications and large amounts of WAN traffic have turned to network optimization tools. For more on WAN optimization, read this tip on SearchNetworking.com.
The accidental green data center
Today, green data centers, energy efficiency and the environment are top of mind for many companies. But just a few years ago, when Panalpina started thinking about a data center consolidation, green computing wasn't even on the radar. Providing the same services with less energy was just a byproduct of the project.
"We didn't do this consolidation to be green," Heinlein said. "In hindsight we're geniuses, but it wasn't about that. We needed to provide a service to our internal and external customers cost-effectively."
At each of the data centers it consolidated, Panalpina had a handful of IBM pSeries servers. All together, the company removed 10 pSeries by consolidating the servers on AIX Logical Partitions, or LPARs. The company also removed four EMC Celera SE network-attached storage devices that functioned as file servers and removed three Exchange clusters. Panalpina also removed eight Citrix servers and four domain controllers.
On servers alone, Heinlein estimates the project saves Panalpina 20% on its data center power bill. And that's not even factoring in the multiple air conditioners, UPSes, generators, and other energy-draining infrastructure removed from the outdated facilities. "Not having to run so many sites is a lot friendlier to the environment," said Bill Fraser, vice president of IT operations at Panalpina. "Instead of having a few servers at each site, you can collapse and scale systems."
Project management takes timing, coordination
According to Heinlein, one of the most difficult aspects of the project was coordinating downtime. Making sure the network is ready, scheduling the migration of users, building up the infrastructure: You have to get the timing right. "At first we moved smaller branches and got a feeling of how many users we could move in one night," said Kevin Ball, Panalpina's network group manager. "We built up experience and got very good at knowing what we could move ahead of time. In the end, we were very efficient."
While that kind of planned migration is very difficult, the Panalpina team faced an even more urgent move when a hurricane struck its Houston data center and Heinlein decided to migrate the site's data center right then based on the company's disaster recovery process.
"When a hurricane hit Houston, we had to shut down the data center and host it out of the New Jersey facility," Heinlein explained. "We'd done a couple [migrations] already, so we transferred the entire operation in 48 hours. The people working for me did an excellent job."
"It was all hands on deck," Fraser said. "But we knew what we could do. It was a fire drill effectively, that came off seamlessly."
Heinlein makes it look easy. But with nearly 30 years in the transportation and logistics business and project management know-how, he has the experience to back him up. "I deal a lot with my internal customers," Heinlein said. "I understand the freight and business processes background they are coming from. We can speak the same language." Heinlein chocked up the success to solid technologies, adrenaline and experience, but he credits primarily Fraser, Ball and the rest of his team for the results.
"I have a great team that helps me do this and advises me," Heinlein said. "They force me to always kick the tires and turn over the stone to see if things can be done better. I say, 'I'm just the guy that writes the checks,'" Heinlein joked.
Matt Stansberry is SearchDataCenter.com's Senior Site Editor. Write to him about your data center concerns at firstname.lastname@example.org.