Issue link: http://cp.revolio.com/i/915954
4 | THE DATA CENTER JOURNAL www.datacenterjournal.com Data CenteR SpenDing LeapS a ccording to a CBRE report, 2017 data center investment in the U.S. screamed higher com- pared with 2016. e company estimated $18.2 billion in 1H17 spending, exceeding spending for the entire previous year. Although the headlines generally go to big-name cloud-service providers, spending also reflected smaller companies investing in on-premises facilities. A Fortune ar- ticle summarized the situation by say- ing, "If this trend continues through year's end, total data center investment in 2017 will surpass the total of the last three years combined, according to CBRE." Partly, this spending boost is to be expected as big cloud providers expand their resources to offer more services. But it also means the outlook for on-premises data centers is less gloomy than it might appear. Living on the eDge Mobile and IoT devices oen bring conflicting demand on resourc- es: ey may require lots of process- ing for certain tasks, yet their small size precludes the kinds of powerful processors necessary to accomplish the task quickly enough to satisfy impatient users. Data centers have the necessary capabilities and can provide them through an Internet connection, but when they're far away, latency can become a concern. One response to this conundrum is to bring the data center (in small chunks) closer to the users, exploiting the benefits of central computing without the latency that long distances create. Edge data cen- ters can be small containers or closets and can reside in almost any location close to numerous users. Although they're nothing new in 2017, they did gain steam. it SpenDing StiLL SLaCk, but the CLouD RoLLS on Worldwide IT spending contin- ues on a plateau according to research firm Gartner. As I mentioned in my year-end roundup for 2016, $3.5 tril- lion seems to be a "magic number" around which global IT spending has hovered for about a decade. And that trend looks to continue; as of October, Gartner predicted $3.5 trillion for 2017 and $3.7 trillion for 2018. Given that recent years have seen regular downward revisions of Gartner IT- spending forecasts, the 2018 number may be optimistic. On the other hand, most patterns break eventually. A chief driver of this flatness in IT spending has been the cloud. Com- puter Economics data, which I cited in the October issue of e Data Center Journal, suggested that declining costs have done a great deal to offset grow- ing demand for resources. And much of that decline owes to IT outsourc- ing through the cloud. As a result, although one might normally expect spending to rise as companies employ more services, falling costs have kept it steady. Moreover, although the cloud isn't the answer to all IT questions, the trend looks to continue for some time. e result may be continuing disap- pointment for Gartner IT spending forecasts. eConomY a-okaY—FoR Some ReaSon M2 money velocity continues to tumble? Check. Civilian labor-force participation rate staying near lows not seen since the 1970s? Check. U.S. federal-government debt still tacking on a trillion dollars or so a year? Check. Economy overdue for a recession, judging from the roughly seven-year cadence over the past six decades? Check. Yet the stock market racks up record aer record, and the headline unemployment number has dropped under 5%, indicating "full employment." It's tough to complain about the lack of a reckoning for the nation's failure to address the problems that caused or have worsened since the Great Recession, but one wonders how long the party will last. In the mean- time, the data center and IT industry can continue to focus on technological and product development rather than whether company bankruptcies and liquidations are in the offing. mooRe'S Law iS DeaD Although Intel has tried to argue otherwise, a growing consensus appears to be that Moore's Law has reached the end of a remarkable run. Among the latest public adherents to this view is Nvidia CEO Jen-Hsun Huang (although he may have market- ing reasons for his position). e end of Moore's Law doesn't mean the end of innovation, but it does mean a slow- er pace of progress in semiconductor manufacturing. Companies will need the enD oF mooRe'S Law DoeSn't mean the enD oF innovation, but it DoeS mean a SLoweR paCe oF pRogReSS in SemiConDuCtoR manuFaCtuRing. CompanieS wiLL neeD to Look beYonD juSt aDDing moRe tRanSiStoRS iF theY want to DeLiveR betteR it pRoDuCtS anD SeRviCeS.