ANDOVER, Mass. -- Data centers are coming up with complex, varied energy demands, whether it's a cluster of servers in a metropolitan high-rise or an expansive server farm on greenfield land in Ohio.
Schneider Electric's IT products revolve around energy management for the data center. The company provides energy management products and services, including StruxureWare data center infrastructure management (DCIM) and various uninterruptible power supply (UPS) systems.
Hal Grant, executive vice president of IT at Schneider Electric, sat down with SearchDataCenter in the company's new headquarters here to discuss energy usage as data centers go global, bigger, and closer to home. In labs and data centers around its campus, Schneider Electric's engineers implement products and reference designs to answer specific needs.
How can IT organizations plan for global dispersion of high-performance data centers?
Hal Grant: The fundamental components of a data center -- the systems and software -- stay the same.
There's one set of experts [at Schneider Electric] who are aware of what's all around the globe, and we learn from the things happening outside of the U.S. It brings better ideas in when you can coordinate that primarily U.S.-based expertise globally. For example, between the U.S. and Singapore, an IT organization supports the same company, with the same needs. But the data center power, land and water use costs are different in each location, and that might influence the facility designs. When you design to use a smaller footprint, that expertise carries over to other sites.
IT organizations with multiple data center locations serve a global function, so operationally they do need to coordinate time zones, standardize metrics, etc. Schneider Electric offers products that track input from various sites on various metrics for one view, or the company performs this tracking as a service.
Are you seeing a bifurcation of the data center industry between Web-scale mega data centers and small edge data centers close to end users?
Grant: Cloud and mega-scale data centers concentrate a lot of compute into a small number of companies. They expect suppliers to customize and optimize products for their needs, which forces change at all levels. When developing products to serve Web-scale data centers [which operate on megawatts of power] you have to use facts to drive the conversation. They're fact-driven rather than tied to legacy ideas. They base decisions on data so they want to collect more data, faster. And the speed of change is accelerated.
The other difference is in the facility itself: Abundant land and electricity, room for renewable energy generation, government support and incentives are all very important to the large data centers. They're driven by finding where it is least expensive to run this data center. You'll find mega data centers growing in rural areas in the U.S. and in Asia.
For data centers located to be in close proximity with end users, network connectivity is a prime driver. You have to consider making investments in uptime if the local power reliability or abundance is a concern. Cooling design matters because you're not always in an optimal climate for data centers. Sometimes, from a facility point of view, you wonder, why did they build there? Colocation and cloud give companies an option to have a server presence in a tough location without building the facility themselves.
After the interview, SearchDataCenter visited Schneider Electric's in-house data center and development and testing data center lab. The 100kW site uses two banks of UPS connected via transfer switches to a diesel generator. Transfer switches ensure a smooth transition off and onto utility power during and after an outage, protecting servers and other sensitive equipment.
How is DCIM evolving?
Grant: DCIM and BMS accomplish different things around the same room. What's important is that they are able to plug into the equipment that they monitor and support.
There's steady growth in DCIM adoption, but it's not shooting through the roof. The important question for IT departments is how do you get value out of DCIM and turn the DCIM data into information?
We visited the labs where Schneider Electric tests the interoperability and integration of building management systems (BMS) with commercial building meters and controls for various real-world deployment scenarios.
In the on-site Schneider Electric data center, smart meters demonstrate this data collection for server power monitoring. Similarly, the company monitors air temperatures to determine whether to use free cooling, evaporative cooling or chilled air from compressors to maintain a steady 74° Fahrenheit in the cool aisle.