Data Center Journal

VOLUME 44 | JUNE 2016

Issue link:

Contents of this Issue


Page 5 of 24

THE DATA CENTER JOURNAL | 3 More Is More—except When It's less C learly, a data center with 99.999% uptime, massive amounts of storage and processing capacity, multiple levels of redundancy, huge network pipes, and the most advanced cooling system initially sounds far better than a small, less reliable, lower-bandwidth facility with modest capabilities. But for a small company with little capital on hand, particularly if its IT services can bear brief periodic outages, the "better" (i.e., techni- cally superior) data center would be a total disaster. Purchasing, let alone maintaining, such a monumental feat of technology would likely crush such a company—es- pecially when something less would do the job just fine. In this case, less is more, and the less advanced data center is the superior choice. But the situation need not be so ex- treme. It might involve narrower decisions, such as whether to expand an existing facility or outsource some resources to the cloud, whether to implement additional power redundancy or just take the risk of a little more downtime each year, and so on. In other words, the situations that compa- nies face in making IT-related decisions don't always come down to a bicycle versus a Ferrari—sometimes they come down to a Toyota versus a Ford. Historically, one might consider the tendency of many data centers to run their facilities at relatively cold temperatures. Doing so ostensibly increases the lifespan of certain electronic systems, but in light of the cost (economic and environmental), the additional infrastructure taking up precious space and other factors affect- ing equipment lifetime, it's impractical. A holistic view of cooling and the data center at large takes more than just technical considerations into account, and the re- sults are sometimes counterintuitive—yet potentially critical to business success. a holIstIc VIeW of data center desIgn e best data center for a given company or purpose isn't just a matter of getting the most technology bang for the buck. Unused capacity, for instance, represents capital that slowly (or quickly) bleeds away through depreciation, on top of extra power and maintenance expenses. A holistic approach considers IT needs and capital, to be sure, but it also takes careful account of what the business requires. e temptation may be to spend the entire allotted budget for a project, but if a company can get what it needs for less, that extra capital can go toward other business-building expenses beyond IT. Moreover, with the growing emphasis on modular data center design, as well as the numerous outsourcing options ranging from colocation to the cloud, companies need not future-proof their facilities for the decades or even just years to come. at freedom to focus on today's needs rather than tomorrow's also frees up capital— sometimes saving tremendously on interest for companies that plan to finance their IT development. Perhaps a company's most criti- cal task, then, is knowing itself—that is, recognizing what it needs and who should have input into that process. Clearly, those involved in delivering the core business know what customers are demanding, but they may not know how to go about meeting those demands on the technical front. IT professionals may know how to meet demands, but don't always recognize the larger business goals. And facilities personnel may understand the infrastruc- ture challenges that everyone faces. Not to be le out, however, are outside consultants and contractors who have experience iden- tifying the right solution for a given project and implementing it smoothly. One aspect of business self-knowl- edge is estimating growth. Company executives recognize that a static business is an uncommon thing: more oen, busi- ness is increasing or decreasing (and both can be scary propositions). Being ready for growth (or even decline) shouldn't be a matter of buying lots of excess inventory and preparing for the worst, but identifying the most likely possibilities and plan- ning an avenue of attack for when those circumstances arise. For instance, in the longer term it's far better to have a strategy for increasing capacity should demand rise than to have lots of excess capacity waiting to become useful (all the while consuming resources). Among the critical considerations is reliability. Aer reading estimates regard- ing the cost of IT downtime, an executive could be forgiven for simply demanding as much reliability as the company can afford. But a high price doesn't always equate to a high value. For example, if $1 million worth of extra infrastructure saves 10 hours of downtime at $10,000 of loss per hour over the lifetime of the equipment, the company has lost $900,000. Estimating losses from downtime is difficult, though, and the numbers may vary depending on the type of event. For instance, 10 hours of downtime spread in small increments throughout a year may be far less costly than two 5-hour-long events, even though the aggregate amount of time is the same. Taking account of potential losses is important, but so is doing so honestly—a mistake either way can be expensive. In addition, the IT industry's growing awareness of the costs of excess capacity can easily result in going to the opposite extreme. e siren call of the cloud is "no up-front costs!"—but that doesn't neces- sarily mean you pay any less in the long term. For many companies, the issue will be finding the right balance of capital and operational expenses rather than all of one versus all of the other. And again, at the heart of that matter is the market and the company's particular needs. conclusIons No two data centers are created equal—nor should they be. And in con- sidering one data center design versus another, far more is involved than mere technical capability, or even cost. Busi- ness needs should drive a company's data center decisions on everything from server capacity to uptime goals to reliance on outsourcing. A holistic design should con- sider input from every part of the business, not just IT. In recognizing that a particular design suited to one situation is likely un- suitable for another, companies will gain tremendous freedom to find the right solu- tion—but exploiting that freedom takes work, particularly in knowing oneself. n

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - VOLUME 44 | JUNE 2016