As is my usual practice for this annual predictions column, I asked the official Hot Aisle Insight crystal ball for guidance. It did not answer, instead, “Alexa” immediately responded with “Model XYS GPS is on sale today for only $49 — just say ‘Buy’ in the next 17 minutes, it could be delivered tomorrow.” While I was not planning to buy a new GPS, it seemed like too good a deal to pass-up, especially since it included free shipping!
This also made me think about testing the recently developed, highly advanced, Hot Aisle Insight Artificial Intelligence engine, which is based on Linguistics Thinking (code name HALT). I asked it to predict the presidential election by analyzing related tweets, since that seems to be one of the primary methods of communicating key political position statements by the presidential candidates. HALT made a deep analysis which took about two minutes, and at first I was impressed with the accuracy of its first response, “There is a 99.876% probability that one of the major party candidates will win the election.”
Clearly, HALT will need more development, so I tried asking my smart watch about the candidates’ positions on data centers and much to my surprise the importance of data centers and their impact on the economy were not mentioned by either candidate, although emails (and their security or lack thereof) and the U.S. economy seemed to be the most frequent subjects.
While I usually try to predict what technology developments and trends will impact the course of data centers for the coming year, I, like most of you, have been amazed at the explosion in data center growth, mergers, acquisitions, and divestures across the spectrum of players, which has become a game of musical chairs. This new paradigm encompasses wholesale and retail colocation providers, telecom carriers as well as hyper-scale cloud services, all of whom cannot seem to build, buy, or lease data centers fast enough to meet demand (or predicted demand) driven by the so called digital economy (what was last year – the analog economy?).
Moreover, while the classic enterprise data center is not dead, fewer organizations are choosing to build, own, and operate their own facilities. They would rather invest in managing their own IT infrastructure to be able to meet the escalating demand for performance and capacity, and leave the operation of the facility to a provider while others never want to own, much less even see a server, and want nothing but cloud. Customer demands are creating rising and highly dynamic requirements for more computing, storage capacity, and of course network bandwidth. Yet, some major telecom providers such as Verizon cannot decide if they should go bigger or sell their data centers.
Moreover, ever changing market conditions have caused senior management to reevaluate where and how to invest their capital in the facility in an effort to optimize and allocate more resources toward their core business. As a result, the strategic decision factors have evolved past the previous focus on just the costs of building, owning, and operating the physical data center.
Even some very sizeable organizations which have the financial resources to build large scale facilities are reconsidering the need to build or expand and own their own fleet of geographically diverse data centers to meet rising demands. They are also reevaluating the burden of adding operational and management staffing at each site, to an already challenging task, due to a shortage of skilled personnel. While retail and wholesale colocation offerings have been available for long time, the scale and scope have changed for wholesale offerings. As a result, there has been a significant shift of larger organizations toward wholesale colocation and cloud based solutions, as well as the advent of the “Connected Campus.”
While previous colocation and cloud providers were competing with each other, in some cases they have become mutually symbiotic via the creation of the Connected Campus, which is a growing trend for cloud providers to be situated within a wholesale colocation campus. This provides the benefit of direct low-latency connectivity to the cloud provider via local campus fiber links at minimal costs. This offers a hybrid strategy of relatively low and consistent costs of wholesale colocation, coupled with only paying for peak computing demands via the “on-demand” flexibility of cloud-based services. Again, by competing providers offering higher performance networking coupled with lower costs — delivering a one-stop alternative solution to the classic enterprise choice of “colo vs cloud.”
In many cases the colocation provider may also have already negotiated tax advantages for the site which would help lower the final cost to the enduser organization. Significant financial investments and/or job commitments allow wholesale colocation providers to either obtain new tax incentives or to qualify for existing programs. Currently, more and more states are offering some form of incentives to bring large scale data center projects into areas that need economic development. At this point in time, more than half the states in the U.S. have varying forms of tax incentives. For example, some states such as Virginia have found that sales tax and energy incentive programs can have a profound effect on data center growth resulting in the massive “Data Center Alley” where the top data center players now develop 100 acre sites, each designed to support several million square feet of whitespace.
From a facilities perspective, even some the classic risk-adverse enterprise customers who consider colocation as an option, have broadened their view on maintaining the traditional 68ºF closed loop cooling in favor of operating warmer temperatures within the “recommended” range (64.4ºF-80.6ºF) of the ASHRAE Thermal Guidelines.
Wholesale colocation providers in particular have become hyper focused on this since they understand this helps make their offerings more economically competitive to their customers. They have also incorporated airside economizers to reduce mechanical cooling energy requirements. And while Google, Facebook, and other hyper-scale businesses have been doing this for many years, it has taken the enterprise manager some time to re-consider airside “free cooling” as an option as long as it helps lower their operating costs. While seeking lower PUEs have become part of the now commonly accepted market yardstick when assessing the facility’s energy efficiency, the new highly contested 2016 ASHRAE 90.4 Energy Efficiency Standard for Data Centers will make it mandatory as it becomes adopted as part of local, state, and federal building codes in the coming years.
So while reliability and availability as well as the presumed number of “9”s have been the cornerstone mantra of the data center culture, we ultimately have to consider “cost per click.” This is something that the internet giants have learned and mastered as they grew over the years and what eBay first revealed in 2013 to the public when they created the digital service efficiency (DSE) dashboard, which publicly disclosed PUE, WUE, CUE, energy used, and, uniquely, transactions per kWh. In effect, it is the Holy Grail for any CFO providing a direct connection between what their customers do on their websites and the total cost of providing services to them. (Note – the DSE dashboard apparently no longer seems to be publically accessible at the time this article published, but the reference is still posted at http://www.ebaytechblog.com/wp-content/uploads/2013/03/FINAL_DSE-Solution-Paper.pdf.)
So as we move into 2017 we can all be expected to look closely at every aspect of TCO because data centers are no longer just for storing and processing data — they are part of the sales and delivery cycle and like any other line item cost in a P/L analysis, it is no longer a “sacred cow” and must deliver value.
While the major large scale efficiency improvements have long been related to reducing data center cooling costs, this year multiple manufacturers of large scale UPS systems have efficiency claims that seem to be knocking on the doorsteps of the “100% Club” (touting 98 to 99% efficiency even while in double conversion mode). While UOS efficiency measures such as “ecomode” (which essentially is a form of static bypass) have been offered for several years it is due to the improvements in silicon-carbide (SiC) technology which substantially and amazingly reduces the total losses in the AC-DC-AC on-line double conversion process. Moreover, several major UPS manufactures also offer lithium ion batteries as an option despite the fact that they are two to three times more expensive than traditional VRLA lead-acid batteries. Why? They are being positioned as having a lower overall TCO since they are expected to last “much” longer than VRLA batteries and take up less space as well as not being as sensitive to higher temperatures. However, conventional VRLA battery manufacturers are also seeing the trend to lower costs and at least one vendor suggests the possibility of 30 seconds of VRLA run-times to save costs (http://www.missioncriticalmagazine.com/articles/88693-who-moved-my-ups).
THE BOTTOM LINE
My data center prediction for 2017 is simple, everyone will need to view every data center design and virtually all related equipment purchases though the TCO looking glass. While cost effectiveness is far from new, every data center equipment manufacturer, and data center builder, is not just competing on initial purchase cost, they are all directly or indirectly discussing TCO as a key part of overall design and marketing strategy.
We have been fully engulfed by the Digital Economy and must remain connected 7x24 by all manner of mobile devices, which now surpasses the PC as the device users seem to use most to view information, communicate, and most importantly make purchases.
In effect we finally have come to realize the true reason the internet exists — the “Buy Now” button. In fact, for those who find even “One Click” ordering a bother, there is “Alexa,” which started in Amazon’s Echo, but is now embedded in all their new Fire TV devices. So if you even “See Something” while you are watching something, you can “Say Something” simply by say “Buy” — just be careful what you say, since Alexa is listening to every word (all the time).
Moving into 2017, we can all be expected to look closely at every aspect of TCO because data centers are no longer just for storing and processing data — they are part of the product, sales, and delivery cycle like any other line-item cost in a profit/loss analysis. It is now part of the bottom line and is no longer a “sacred cow” and must become a “cash cow” expected to deliver continuously improved performance at the lowest TCO in order to remain competitive in the Digital Economy.
Finally, speaking of competitive, since this is the time of the year where shopping is a make or break event for many businesses, start shopping and stay tuned to see how these predictions pan out. Have a happy, semi-secure shopping experience, be it in-store, on your favorite mobile device (or by speaking to Alexa, Cortina, Google Assistant, or Siri), and have a Green Holiday Season and a Happy Sustainable New Year! n