This audio was created using Microsoft Azure Speech Services
Most of us are familiar with the new demands impacting information technology (IT) management and data center physical infrastructure (DCPI) layers, as detailed in my last post. To better adapt to these changes, it’s time to think about ways to apply data science to the way we manage data centers.
Data science can be defined as the extraction of useful information from a data set. Unlike traditional business intelligence, it’s not backward looking nor tied to a centralized data mart, it can bring together many distributed sources, perform modeling and analysis, and use that to deliver predictive, actionable insights.
While data scientists typically work with computers to perform trending, regression, and clustering analyses, ironically, one area data science has not been applied, at least up until recently, is the fine-tuning of data centers and DCPI.
Rightly applied, data science allows a data center to predict and adopt to changes in the IT load and new demands coming from the business system and consumer layers of IT. Dynamic planning, dynamic provisioning, rapid adjustments to power and cooling assets, and ensuring the DCPI is able to handle increasing kilowatts (KW) per square meter (SqM) are some of the things that data science can be used for. It all comes down to being able to adapt to dynamic changes driven by the consumer layer without overbuilding or running into problems with performance or availability.
Sounds good, but if it can address these key issues, then why isn’t everyone doing it? The short answer is because it’s hard, and few companies possess the skill sets to deliver it. According to research from McKinsey Global Institute, in the U.S. market alone, there is a shortage of 140,000 to 190,000 people with the deep analytical skills needed to extract insights from Big Data, and a lack of about 1.5 million other managers and analysts with the skills to understand and make decisions based on the analysis of big data.
The application of data science involves more than analysis of structured data. With data centers, there also is vital unstructured data such as energy audit or field service reports, or third-party data on weather, than might need to be weaved into an analysis. So the pure data scientists should be collaborating with experts in field service, data center design, and use of data center infrastructure management (DCIM) software to ensure all needed information is incorporated into the analysis.
As Hilary Mason and Chris Wiggins, two experts and authors on data science, have stated regarding the diverse talents involved: “data science is clearly a blend of the hackers’ arts, statistics, and machine learning … and expertise in mathematics and the domain of the data for the analysis to be interpretable. It requires creative decisions and open-mindedness in a scientific context.”
So at the end of day, how many companies operating data centers really have the expertise to start applying data science, or the time, given the daily operational tasks involved with running a data center? It’s a safe bet to say that very few are positioned for it. Schneider Electric has moved to address this gap by bringing on board data scientists who can collaborate with DCPI experts and services managers to analyze data and unstructured content, make predictions, and adapt as needed. So while most companies probably don’t have the needed expertise, they can bring it on board as a service.
Here’s a few questions that be used to gauge whether a company should consider a data sciences approach to data center infrastructure optimization:
- Are you monitoring your environment effectively?
- Can you predict asset performance?
- Are you committed to operational efficiency and continual improvement in your facility?
- Do you have the time and expertise to both operate and optimize your critical facility?
- Can you evaluate and optimize your infrastructure over the entire lifecycle?
If your answer was in the negative for most of these questions, a new approach to data center improvement is likely in order, leveraging data science as a way of adapting your infrastructure to seemingly unpredictable demands.
In a following post, we can look further at the capabilities needed as part of a services offering that applies data sciences to data center infrastructure improvement, but hopefully, you now have a better understanding of what data science is, the skills gap for data scientists, and why data science as a service holds potential for enabling data-driven decision making and continuous improvement for the data center.
