Data Center Journal

VOLUME 54 | FEBRUARY 2018

Issue link: https://cp.revolio.com/i/938019

Contents of this Issue

Navigation

Page 9 of 22

THE DATA CENTER JOURNAL | 7 www.datacenterjournal.com meet customer demand, or should it governments impose it through taxes, regulations or both? In the first case, efficiency increases will come if cus- tomers demand them; assuming en- ergy efficiency means cost efficiency as well, then these increases will be an uncoerced byproduct of market forces (i.e., competition). at seems to be the current situation given the rela- tively light tax and regulatory burden on data centers. Taxation and regula- tion, although perhaps well meaning, would likely reduce competition by further centralizing resources in the hands of large companies that can more easily bear the brunt of these government interventions compared with small or midsize companies. Furthermore, customers would suffer from lesser competition, and so might the natural market drive to greater efficiency. Rather than innovating, companies might simply aim to meet government standards regardless of whether they represent the best pos- sible result. At the heart of the third ques- tion, is efficiency an unadulterated blessing, lies the Jevons paradox: an observation that increasing efficiency tends to increase, rather than de- crease, consumption. IT has certainly exhibited this trait; efficiency has greatly increased if the LBNL report is correct, but so has consumption of resources. e result has been a slow rate of increase in data center energy consumption over recent years—certainly slower than some gloomy forecasts predicted, but not the decline that some may prefer. Yet the increase seems to be similar to that of the broader economy, albeit at a slightly faster rate (unsurprising for a still-developing technology). e Jevons paradox, however, doesn't mean efficiency is a bad thing in this context; it simply means demand will respond to greater ef- ficiency (if it means lower prices) by increasing. Without the greater efficiency, the result might have been similar energy consumption by data centers anyway—but with far, far less to show for it. e salient issue then becomes whether government should both mandate efficiency and suppress demand through taxes and/or regula- tions to beat the Jevons paradox and reduce energy usage. LIMIT DATA CENTER ENERGY CONSUMPTION? e notion that increasing data center energy consumption is neces- sarily bad assumes the increase isn't balanced by changes elsewhere in the economy. For instance, per- haps the proliferation of computing resources means more workers can telecommute, cutting down more than proportionally the consump- tion of gas. Although a quantitative analysis would be difficult in this case, recognizing these balancing factors is critical. And it's in light of such con- siderations that one should weigh government intervention in the data center market. As I mentioned above, regulations may increase efficiency, but they also hamper competition and could therefore cause more harm than good. Taxes are in the same boat. Yet the temptation is to employ taxes and regulations to remove the driving force behind the Jevons paradox: they can lead to greater efficiency while tamping down demand, potentially reducing energy consumption where- as the market, le to itself, would increase energy consumption. Given the relatively slow increase of data center electricity consumption over recent years (specifically, com- pared with the early 2000s) and the negative side effects of taxation and regulation (just look at the health- care industry), leaving the industry in the hands of market forces is the best approach at this point. Data centers today are far from gobbling up a giant portion of U.S. electricity production. By comparison, data center electric- ity consumption roughly doubled from 2000 to 2006, reaching 60 billion kilowatt-hours. In the time since (at least to 2014—almost a decade—us- ing LBNL numbers), it has barely topped 70 billion kilowatt-hours. e case for alarmism is weak, at best. CONCLUSIONS e drivers of increasing data center efficiency remain in play despite the relative absence of gov- ernment intervention. For example, technologies such as more-efficient cooling, virtualization and non- mechanical storage (SSDs) are still diffusing and/or falling in price. Moreover, service models such as the cloud are centralizing resources and, because of their scale, increasing efficiency. Market forces—the collec- tive interaction between data center companies and their customers—are making each unit of IT resources con- sume far less electricity year by year. Although total electricity consump- tion is probably still rising, the rate is by no means alarming and is similar to that of the broader economy. Before looking to taxes and regulations to accelerate efficiency improvements or to artificially cut electricity consumption, we need only consider the health-care industry: a monstrosity of regulations and com- plexity whose prices have increased far faster than the U.S. GDP. IT, on the other hand, continually offers more for less cost—both in price and energy. Striving for greater efficiency and better environmental stewardship are important, but in the case of data centers, the best approach for further development in these areas may simply be to trust that companies and their customers will work it out. n

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - VOLUME 54 | FEBRUARY 2018