Time for IT and Facilities to Talk
by Danny Bradbury
If Dave Eastman hadn’t become a leading voice in the data center community, we get the feeling he may have been a marriage counselor instead. Over his near-20 year career in data center management, he has spent a lot of time sorting out communication problems between two specific kinds of technology professional that just can’t seem to get along.
Eastman worked for Broadcom between 2000 and 2015, and was asked to bridge the gap between two notoriously misaligned parties: IT and facilities. He will be discussing how he did it, and some of the problems that the two teams encounter on Wednesday, March 14, at his Data Center World 2018 talk.
“IT has one interest, and that’s keeping their servers and their applications running,” says Eastman, who is now the VP of colocation and data center management company server farm’s InCommand IT infrastructure management solution. “Whether the data center operates efficiently or not is no concern of theirs.”
That’s partly down to budget structures, he says. It is typically the facilities budget, rather than the IT budget, that pays for the data center. However, it is also a function of organizational structure and how people work.
“No one IT person is in charge of deploying servers and applications in the data center,” he says. “Depending on the size of the organization, it could be 10, 20 or 50 people that deploy applications in the data center and they don’t collaborate well. They are all individuals working in their own IT silos.”
This problem is compounded by a lack of tools on the facility side. Many facilities teams are still under resourced when managing power and cooling, he says, often still recording readings manually with clipboards. This makes it difficult to understand what is happening on the IT side and how it affects facilities. The first time the facilities team may be aware of a problem is when alerts start going off.
This disconnect creates some problems for data center operators. Often, IT people fail to tell facilities before they install servers in a rack. When the alarms start going off, the facilities team must pick up the pieces or tell the IT department to roll back its changes.
Even when the two sides do talk to each other, things can still go awry due to a lack of process. Eastman recalls one incident when an IT team told facilities that it wanted to deploy 10 new racks of servers, and got permission.
The IT team went through the budget cycle and deployed the servers six months later. Alarms went off, because facilities didn’t reserve the infrastructure or record the request. IT may have obtained permission, but by the time it deployed the servers, everything had changed.
There are ways to solve these problems, Eastman says. The first involves creating toolsets that make it easier to gather data about the infrastructure and present it easily to the facilities team so that they can plan for changes more proactively.
The second involves bringing IT and facilities closer together by making organizational changes. IT must become more organized and less siloed when deploying equipment and applications in the data center, he says. To help solve this problem, he developed a new organization in the IT group that managed its deployments in the data center, providing a single point of interaction for facilities.
“You can’t have 20 or 50 people making change orders,” he says. “There can’t be two separate teams to run the data center. There has to be one team, using the same set of tools.” In fact, he says that in an ideal situation the IT department will assign some of its own team members to work with facilities on a full-time basis.
Finally, he advises data center owners to ensure that decisions made by this integrated team stick. They can do that by introducing mature change management processes. “If a company does not have good change control processes, then it’s dead on arrival,” he says. “They will never be able to optimize and run an efficient data center.”
Good change management thrives on consistency, he concludes. Making changes in the data center more process-driven than exception-driven will make operations smoother and throw up less challenges for teams as they find harmonious ways to work together. Just as in a good marriage, everyone is happier when everyone communicates, and everyone knows what to expect.