Volatile IT layer demands drive a data science approach to managing critical infrastructure

This audio was created using Microsoft Azure Speech Services

All those nice apps on all those smart phones around the world are great for consumers in that people are able to take care of their shopping, bill payments, banking, entertainment, and many other needs at any time or any place. But all that activity coming from the consumer layer creates enormous demand on the supporting business system, IT, and infrastructure layers

Those of us involved with data centers solutions are well aware of this dynamic. The old days of data center demands being more predictable, driven largely by corporate computing needs, has given way to a much more volatile environment in which digital commerce, mobile broadband, and smart, connected devices have led to explosive and unpredictable demands on the IT layer comprised of servers, virtualization, and other technology, as well as the “infrastructure” layer comprised of racks, cooling, power protection, and other data center physical infrastructure (DCPI) components.

The proof points behind these shifts is all around us. For example, according to IDC, smart phone shipments outpaced feature phone shipments worldwide for the first time in 2013. Meanwhile, Cisco Systems estimates that as of 2012, there were 8.7 billion devices hooked into the “Internet of Things,” and that by 2015, there will be 15 billion devices connected. These types of changes are worldwide, but are especially abrupt in rapidly developing economies in Asia Pacific where we have a growing middle class and rapid adoption of new technologies.

On a more micro level, if a bank launches a new mobile app today, or if a consumer electronics company launches a new “smart,” connected product, the demands on their IT and infrastructure layers would change almost immediately.

The bottom line is that the demands on the IT and infrastructure layers are massive, volatile, and thus hard to predict and understand. At the same time, there is also a skills shortage. According to a survey from Data Center Dynamics, 60.5 percent of respondents named the data center skills shortage as a key area of concern, ranking only behind reduced budgets and rising power costs. For the 10 countries that comprise the Association of Southeast Asian Nations (ASEAN), figures show that the region possesses only about 5 percent of Uptime Institute accredited data center design engineers in the world, even though data center capacity for the region is expected to grow 40 percent to 50 percent over the next five years.

But for many companies which operate their own data centers, and even for tech companies or colocation providers for whom data centers are a core focus, simply keeping data centers up and running is a huge challenge. Few companies have the in-house expertise to fully monitor, understand, predict and adapt to the demands now being placed on the IT layer and the DCPI when they are so busy with daily tasks like changing out racks, provisioning servers, or maintaining DCPI components.

So what’s needed? What’s the best response to this mix of explosive, volatile demand from the consumer layer, and ongoing skills challenges in IT? There are many aspects to the complete answer, but essentially, the market must apply better data science to management of the infrastructure layer to understand what has happened, what is happening, and what will happen. Data centers should be able to scale, evolve and adapt, while also ensuring that they maximize the value of their assets.

In a follow-on post, I’ll more fully explain what data sciences is and how it can be applied to data center infrastructure, but a short definition is that data science is about the extraction of useful information from a data set. It is normally a combination of computing, mathematics and know how, but to apply it to DCPI, there also is the need for domain expertise, data center infrastructure management (DCIM) software to aid with modeling and analysis, and the integration of documents and other unstructured data, such as audit or site visit reports, into the analyses.

Stay tuned for more on what data science can do for data centers, and how to apply data sciences under a services offering.

Tags: ,