Download the authoritative guide: Cloud Computing 2018: Using the Cloud to Transform Your Business
For a company to have its own data center is surely a sign of success -- and also a major cost burden. Kenneth Hess discusses pathways to more efficient data center operation.
Data centers are energy vacuums. Historically, they are sinkholes for money, power and inefficiencies. They don't have to be. If you're thinking of building a new data center or if you already have one "on the slab," you can pick up some pointers from the company that knows how to build an efficient data center: Google. These five data center optimizations are those that Google follows for its data centers.
1. Location, Location, Location
Does Google have a 100,000 square foot data center in Silicon Valley? No. Why? It's too expensive. Land is expensive. Power is expensive. And, people are too expensive. It's impressive to have a nice data center in Silicon Valley, but it does mean it might never turn a profit. Try moving your pride to a better, less expensive location. Google's U.S. data center locations: Berkeley County, South Carolina; Council Bluffs, Iowa; Lenoir, North Carolina; Mayes County, Oklahoma; and the Dalles, Oregon. All are a far piece from the pomp, circumstance, and expense of Silicon Valley, but is your company's stock price sitting at more than $500 per share? Google's is.
2. Manage Air Flow
Google states that, "Good air flow management is fundamental to efficient data center operation," and you should, "Start with minimizing hot and cold air mixing and eliminate hot spots." Data center air flow management is far more important than cooling and can, in fact, lower your need for additional cooling. One method for reducing the mixing of hot and cold air is to arrange your racked servers in a hot-aisle, cold-aisle configuration. In this scenario, cold air blows on the front sides of server systems, while the hot aisle's heat dissipates via exhaust fans. Adequate air flow also eliminates hot spots.
Read the rest about data center efficiency at ServerWatch.