Data Center Journal

VOLUME 50 | JUNE 2017

Issue link: http://cp.revolio.com/i/833286

Contents of this Issue

Navigation

Page 9 of 32

THE DATA CENTER JOURNAL | 7 www.datacenterjournal.com o ne size fits all is great for hats but not for data center infrastruc- ture. Time and again, IT leaders subscribe to a data center model that uses an out- dated, one-size-fits-all approach to ap- plication management. For emerging technologies to take root and thrive, these leaders must challenge the status quo and incorporate forward-thinking infrastructure and platform solutions that are less expensive, are more reli- able and allow for scale exceeding that of traditional application-deployment regimes. Legacy is a powerful influence on the way we do things, be it how we go to work or how the IT community approaches a problem. Habits bring us all comfort and assurance that the job will get done and businesses will accomplish their mission. Consider the central data center model, a legacy dating back more than half a century to when comput- ers were expensive machines filling rooms and requiring a large staff to keep them running and fed with batch jobs. As computers became smaller, the staff to serve them continued to grow, and data centers added more resources to meet the expanding role of IT in businesses. e data center became a fortress. Executives became comfortable with the idea of building and maintaining this fortress, both figuratively and literally. e fear of risks from the outside world was al- layed by central IT operation within the corporation's boundaries. e fortress engendered comfort in the physical protection of the company's most important secrets. But fortresses are difficult to expand. e IT fortress was built for a specific purposeā€”to run in a consis- tent fashion without drama. It doesn't handle experimentation or change well, be it sandboxing a new appli- cation, scaling for seasonal surges in product orders or handling the requirements of high-performance computing. And for companies look- ing to expand IT's value, a familiar feeling has re-emerged and is forcing change: fear. e fear of falling behind and of failure is driving IT to look beyond the fortress model for faster, flexible solutions. Traditional assumptions about what must reside inside the walls of the IT fortress have been challenged. As technologies emerge, corporate leadership's assumptions of comfort and safety are tested. Busi- ness leaders know that applications now define their ability to develop their business. IT's mandate is now to manage the risk and move beyond the mentality of the fortress. New applications and functions, such as research and development (R&D), are typically the easiest to move out of the IT fortress. R&D is owned by the early adopters of new technology that may not be ready for production in a business environ- ment. ey can take a more aggressive posture on risk and push the bounds of the possible. R&D teams should experiment beyond the fortress, look- ing for infrastructure and platform solutions that will generate better returns on investment once they move into production. Applications that must scale up and down on a predict- able basis also are prime candidates. CFOs especially want to avoid paying for resources that sit unused, so any function that surges up and down is a prime candidate for moving beyond the fortress and into a solution that can scale up quickly and scale back down just as quickly to save cost. By engineering applications that remain secure in a shared, elastic environ- ment, the IT organization can drive savings while increasing revenue. But the biggest applications that need to live beyond the IT fortress are also the biggest and fastest growing in compute power. Machine learning is a "hard" problem, requiring a brute- force approach for many applications and taxing compute and storage resources. Even "simple" AI functions available through APIs today, such as voice and facial recognition, must go through an intense training period to transform the underlying deep neural networks into functional tools that drive better learning. As a new application for many organizations, machine learning is a perfect candidate for moving beyond the fortress and into an optimized compute environment. Few enterprise data centers are built to the standard that the latest in machine-learning technology requires. And that tech- nology is changing rapidly. By looking beyond the IT fortress, business leaders can ensure their capital isn't stranded in yesterday's technology. But with any application, secu- rity is a major concern. e fear of a breach can cripple the decision mak- ing of an otherwise agile executive team. But hiding inside the IT fortress is no solution to these fears. Any or- ganization, regardless of size, that has network connections to the rest of the world is at risk. Numerous security breaches over the years have dem- onstrated that the IT fortress is only as impenetrable as the processes it contains. And repeatedly, we see that processes are insufficient to prevent the next security breach. Ironically, projects that move beyond the IT fortress may be less vulnerable owing to the scrutiny they receive in moving from the corporate infrastructure. e

Articles in this issue

Links on this page

Archives of this issue

view archives of Data Center Journal - VOLUME 50 | JUNE 2017