Issue link: https://cp.revolio.com/i/938019
4 | THE DATA CENTER JOURNAL www.datacenterjournal.com B ut just when an exponen- tial curve becomes ter- rifying in its implications, it oen levels off—just as world population seems to be doing. e key is to look at the factors influencing growth, whether negatively or positively. In the case of world population, the plummeting fertility rates around the world may put a strong cap on growth, possibly leading to population declines. But what factors are affecting data center energy consumption, and what—if anything—should be done about it? Let's take a look. DATA CENTER ENERGY: NOT AS BAD AS IT SEEMS Similar to Ehrlich on the world- population stage, the data center industry has its own grim prophets. In 2016, for example, Peter Judge wrote an article titled "e Truth Is: Data Center Power Is Out of Control" at DatacenterDynamics. He cited electricity-usage numbers that Google reported in 2011 and 2015, saying that the search giant's "total power usage seems to have gone up 12-fold in the last four years. at is not far off dou- bling each year." He then speculated that similar trends among other major IT companies mean growth of elec- tricity consumption by data centers worldwide. at same year, however, Law- rence Berkeley National Laboratory (LBNL) issued its United States Data Center Energy Usage Report. e report estimated the energy consump- tion of U.S. data centers from 2000 to 2014 and provided forecasts to 2020 on the basis of various assumptions about technology improvements and data center management prac- tices. e authors said that "electric- ity consumption of data centers has been relatively flat in recent years," a perhaps unexpected trend that's "at- tributable to many factors." ey also noted, "Growth in data center energy consumption has slowed drastically since the previous decade. However, demand for computations and the amount of productivity performed by data centers continues to rise at substantial rates." In one case, the report's au- thors charted data center electricity consumption in their historical and forecast range (2000–2020) compared with estimates of electricity consump- tion for a 2010-based energy-efficien- cy scenario. at scenario assumed, for instance, constant industry-wide server utilization of 14%, an industry- wide weighted-average PUE of 1.89 and so on. (See the report for further details of this scenario.) Figure 1 il- lustrates their findings, with energy savings relative to the 2010 scenario broken down by equipment type. A main point of this figure is that over the historical period (2000– 2014), data center usage topped out near 70 billion kilowatt-hours before declining slightly and then remaining fairly even, with a relatively small rate of increase. e dip aer the accelerat- ing growth of the mid-2000s is likely due to the Great Recession in 2009, but the growth rate appeared to be lev- eling anyway, and it has been relatively low since. For comparison, the U.S. Energy Information Administration (EIA) shows a similar albeit less dramatic trend in overall electricity use. e 2000s exhibited a mostly steady rise with some leveling in the 2010s, as Figure 2 illustrates. e dips in 2001 and 2009 are likely a consequence of the dot-com bust and the Great Reces- sion, respectively. By these figures, data centers account for slightly less than 2% of total electricity end use in the U.S. Worldwide figures may vary but are probably similar in other developed nations. Figure 1. Total data center electricity consumption, 2000–2014 with forecast for 2015–2010. (Courtesy of LBNL, United States Data Center Energy Usage Report.)