Issue link: https://cp.revolio.com/i/860140
THE DATA CENTER JOURNAL | 27 Uptime is important to every industry. No business owner, manager or leader wants to close their doors, shut down production or stop shipping for any reason. Nevertheless, the data center industry is perhaps even more sensitive to downtime than many others. Ours is an era in which the need to access data is growing at a dramatic pace. Cloud computing, for instance, is experiencing tremendous growth. Earlier this year, Gartner projected the worldwide public-cloud services market will grow 18 percent in 2017 to $246.8 billion and will reach $383.4 billion by 2020. That means data center outages are becoming more costly. A 2016 report by the Ponemon Institute indicated that the average cost of a data center outage grew from a little over $500,000 in 2010 to about $740,000 in 2016—a 38 percent jump. u ptime is important to every industry. No busi- ness owner, manager or leader wants to close their doors, shut down production or stop shipping for any reason. Nevertheless, the data center industry is perhaps even more sensi- tive to downtime than many others. Ours is an era in which the need to access data is growing at a dramatic pace. Cloud computing, for instance, is experiencing tremendous growth. Earlier this year, Gartner projected the worldwide public-cloud ser- vices market will grow 18 percent in 2017 to $246.8 billion and will reach $383.4 billion by 2020. at means data center outages are becoming more costly. A 2016 report by the Ponemon Institute indicated that the average cost of a data center outage grew from a little over $500,000 in 2010 to about $740,000 in 2016—a 38 percent jump. For these reasons, building reliability into a data center is top of mind, and part of it is planning for emergency power. e reliability of utility power continues to be an issue in many parts of the country. e American Society of Civil Engineers (ASCE) once again issued a D+ grade to the U.S. energy infrastructure in its 2017 infrastructure report card—a grade that hasn't risen above a D+ since 2001. It said that "without greater attention to aging equipment, capacity bottlenecks and increased demand, as well as increasing storm and climate impacts, Americans will likely experience longer and more frequent power interruptions." Data centers currently plan for power outages in a few ways. Backup power is very important, and for many, that means a single large genset designed to step in when utility power fails. Bridging the gap be- tween the utility and generator power supply (somewhere between 10 and 30 seconds) is typically an uninter- ruptible power supply (UPS). For some data centers, UPS systems with relatively larger running times (on the order of a couple of hours) are the only backup power supply on site. Some even plan for power outages by simply accounting for a certain "ac- ceptable" level of failure. ese approaches aren't neces- sarily wrong, but they are flawed. For example, a single large genset meets the Uptime Institute's Tier II require- ments, and although companies that are only protecting their own data might be comfortable with that level, it's typically unacceptable to data cen- ter customers. If the genset should be out of service for routine maintenance when a power outage strikes—or, worse, if it should experience some catastrophic failure during a power outage—its only backup is a UPS system. Furthermore, UPS systems are intended to provide power between the time a utility fails and the emer- gency system kicks in; they're not intended to provide backup power on their own. Even if they're large enough to provide backup to an emergency generator, getting the generator ser- viced and running before the UPS is exhausted may be impossible, espe- cially in severe weather conditions. In any case, even the largest UPS systems cannot match an emergency generator for sheer running time. In a Hurricane Sandy–type event, data centers need www.datacenterjournal.com