Tennison said Union Pacific has a history of being technologically advanced. Not only did it blaze trails with the mainframe four decades ago, but it also is bringing technology into the engines, which now are equipped with onboard computers, GPS and satellite communications capabilities, he said.
So it made sense that when officials saw that technological demands were changing, they were ready to make the significant investment to make such a shift in their data centers.
For example, Union Pacific officials saw that their mainframe programmers were getting older and that the vast majority of students coming out of college were trained on newer technologies and languages.
"When we brought the [mainframe] in here in the 1960s, a lot of the people [working on it now] came in with it," Tennison said. "It's getting harder and harder to find the right people to work on it. It's a whole lot easier to find people skilled in new languages."
IBM, CA, BMC Software and Unisys, among others, have worked hard over the past several years to attract younger programmers to the mainframe platform. In addition, IBM in particular has been aggressive in bringing new workloads-including Linux and Java-onto the big iron systems
For example, IBM Aug. 14 rolled out a new System z offering that includes seven integrated hardware, software and services packages for deployment of such enterprise workloads as data warehousing, risk mitigation and disaster recovery. Big Blue also unveiled programs designed to entice Sun Microsystems and Hewlett-Packard customers to IBM's Linux mainframe platform.
Tennison said Union Pacific officials first began talking about moving to a distributed system four years ago, and built an engineering prototype designed to prove that the infrastructure could scale and handle the transaction workloads it would need to run.
Tennison and his staff also needed to convince Union Pacific officials that such a major project made sense for the railroad, a process that "took quite awhile," he said.
Many data center technology vendors have services programs designed to help businesses plan such major projects. At an event in HP's Marlborough, Mass., offices in July, officials with the company's data center transformation solutions group spoke about the steps they take with clients as they plan out major data center projects. A key one is getting buy-in from both business executives and internal IT staff members.
Eventually both sides fell in step, he said. For executives, "the biggest thing was really showing them where we wanted to get," Tennison said. A key for the IT staff was getting them the training they needed to be able to work effectively in the new environment, and to put in place recurring training programs to keep the IT folks up to speed.
He said he is designing the distributed environment to ensure that such issues as management and security don't hamper the new platform. For example, Tennison is insisting that all IT people use the same tools and software.
"That eliminates a ton of management headaches that typically come with distributed environments," he said.
Tennison said he and others in IT saw the promise of such infrastructures several years ago, and much of the technology already was in place. However, it wasn't until companies like Google, Yahoo and Microsoft began building out their massive server farms that companies began moving in that direction in larger numbers. It also took SOA vendors awhile to move from talking about the model to actually offering solid products in that area.
"We had understood that we could do this," Tennison said.
Now the railroad is five years away from completing that vision.