Best Practices in Building a New Data Center
How to avoid the pitfalls that many experience in mission-critical facilities
by Barbara Morris, Editor, The DCIM Advisory
Peter Sacco, founder and CEO of PTS Data Center Solutions and partner with nlyte Software, remembers when one of his business school professors said, “When you build a company, don’t try to be all things to all people.“ Sacco laughs and adds, “So what did I do? I went out and built a company that tries to be all things to all people.”
Sacco sees PTS as addressing a fundamental challenge common to data centers today: IT advancements are moving much faster than facilities can adapt. While the large data center operators know where to go — the best and brightest engineering and construction management companies — Sacco understands that the small-to-medium-sized operators are struggling. He says, “What the smaller operator doesn’t have is people, so he or she wants one guy who can manage as much of the project from beginning to end as possible.” With that in mind, PTS provides turnkey IT and facility-based services to narrow the scope, schedule and budget for a project, through design, construction, IT implementation, commissioning, and maintenance and monitoring going forward.
Sacco personally visits 50 to 60 data centers of every size worldwide each year, and including his staff’s work sees the information from over 100 data centers yearly. His experience in and knowledge of data centers is vast and complex, but he breaks it down into four main points to consider:
- Defining data center management depends on your point of view.
- Data center challenges continue to be the lack of space, power and cooling
- Mission-critical facilities have a finite life span.
- A data center infrastructure management (DCIM) solution simplifies data center management.
Defining data center management depends on your point of view.
In Sacco’s view of data center management, a facilities manager understands a data center is all about space, power, cooling and physical security. For the IT infrastructure manager, it’s servers, routers, switches and firewalls. For the applications manager, it’s all about database connectivity and applications. But a data center is all of those things and more.
Data center challenges continue to be the lack of space, power and cooling.
As IT demand continues to outpace the ability of facilities to adapt to it, facilities cannot adequately plan for future demands nor adapt quickly to the current environment. Sacco says that he has surveyed data center managers he works with, and 80 percent of them say they’ll run out of space, power and/or cooling within two years.
But if he were to ask the same group what the causes are, they would all say the same thing, “I’m deploying blade servers; I’m virtualizing.” Sacco believes that many data center managers think that virtualization saves energy because it compresses functionality down into a limited number of servers. He adds, “We’re putting more densely-powered servers into the empty spaces, with more power and cooling demands. So we overtax our facilities; that’s why we’re running out of space, power and cooling.”
Mission-critical facilities have a finite life span.
Today, data center operators face not only overtaxed facilities, but also older buildings that suck power and energy. Because data centers have a finite life span, the management team needs to make a decision. They can use a collocation facility, upgrade existing facilities or build a new data center.
Sacco says that the number one consulting task he faces is the client who calls and says, “Hi, I have a data center somewhere in the world. I’m running out of space, power and cooling, and I need to know what to do. Should I fix it in place, build it brand new somewhere else or go co-location?”
Sacco has an answer for each of these questions. First, he believes that once a data center reaches its maximum capacity, it’s better to build new space next to it with additional capacity. “Because data center managers are inherently risk averse, doing an upgrade of a data center in place may not be the best idea; it’s a fairly chancy proposition.”
He continues, “Once you build a data center to be a maximum of something, it’s almost always better to build space next to it with additional capacity, rather than try to modify the existing space to increase capacity.”
Regarding co-location, it may work for some but not for others. Sacco says, “Ultimately, it will always cost more to put my equipment in someone else’s space than to build my own space. Since a data center manager is in the business of making money, which means a 40 percent gross margin, he’ll pay a premium to be in someone else’s data center, probably about 40 percent, so it might not be a logical choice.”
A DCIM solution simplifies data center management.
Data center managers need to track the physical location of every asset in the data center for both IT and facilities. The facilities manager needs to understand the interdependencies of the infrastructure, the power draw for every part of the infrastructure, the power/cooling it consumes, the space it takes up, what type of management resources it consumes and how operators manager capacity planning for all of that.
Sacco says, “From the IT perspective, the manager sees that he’s running out of memory, disk capacity, RAM, etc. In that case, capacity management bridges the gap between IT and facilities to understand how to manage energy consumption, as well as the connectivity and management of all the devices.”
A data center infrastructure management (DCIM) solution, such as the DCIM solution that nlyte Software has developed can provide valuable data for decision-making in this type of situation.
What is DCIM? DCIM means different things to people in the data center, depending on what their needs are. Sacco believes that nlyte is doing a very good job of understanding the entire industry, as well as providing specific pieces of DCIM to address those needs. He sees what nlyte does best as:
- Understanding the physical layout of the equipment by rack elevation (layout of data center and layout of equipment in the rack).
- Providing a monitoring capability so that managers can correlate near real-time data to physical data.
- Offering a workflow solution such that if a manager is going to make changes to the physical layout, he can have a process around how to make that change.
- Providing a certain degree of capacity management such that a manager can make determinations on how much space, power and cooling are available.
“The nlyte product does all of those things particularly well at a very good price as compared to the competition,” Sacco adds. “nlyte doesn’t do this simplistically but in a simplified way. It is helpful in making future projections and is the best in the industry at locating assets and correlating them to real-time monitoring.”
He concludes, “Although there are many tools, working with a third-party solution like nlyte is the best path to success. nlyte has gone through our rigorous exercise of determining which software is the best and nlyte is the best solution.”