I’ve been getting a lot of questions lately about Winter weather, lightning protection, and severe weather in general. Brought about by the approaching Winter season, these questions are usually in the context of what should a data center be doing to protect itself.
With all the planning that one goes through in building or collocating in a data center, the one thing you can never change after the fact is the location. Business drivers usually dictate that a data center must be located in one particular area or another. Knowing the location, there are a number of industry frameworks that address proximity risks for data center location (e.g., BICSI-002), but geography and weather are regional factors. Given your location, these are things one must live with.
When data centers are constructed, they are sometimes built as a greenfield new building, and sometimes within an existing building shell. Local codes and regulations often drive construction requirements for mission critical buildings. For example, seismic bracing in areas prone to earthquakes, wind velocity requirements in areas prone to wind storms.
Lightning protection is something that’s necessary everywhere. A brief moment in time when lightning strikes could mean significant loss for the Business. I’ve personally been involved with data centers which have had to recover from a lightning strike, and it is always a memorable experience (and not a good one). Local codes and regulations have requirements for lightning protection. I know that these are sometimes “value engineered” out of the implementation, but this accepts a serious risk as a result.
Weather patterns are seasonal, so we have to look at the data center in the context of a full year’s cycle of seasons, and risk planning for the effects of storms are fundamental in good data center planning.
When we think of severe storms, the most common threat is loss of utility power, or flooding. We know how to apply backup generators to supply power when the utility goes out, but flooding is often not given enough credit. An example is Super Storm Sandy, which hit Lower Manhattan a few years ago. Lower Manhattan has a large concentration of data centers because of the financial district located there. Diesel generators located on the roofs of buildings were available, but the flooding caused the pumps for basement diesel fuel storage to fail, resulting in extended downtime for those data centers.
When we create contingency plans for severe weather, one often consults the local weather historical data. It’s important during this design step, to consider that changing weather patterns across the world today are causing severe storms to be either more frequent or more intense. Hundred year flood plains can be 25 year floodplains now. That recognition should go into severe weather contingency planning.