The demand on data centers has never been higher, and it’s only going to grow. How do we build and run today — for tomorrow? The Data Center Tech Bytes podcast features big industry conversations that take place in just a small amount of time. Join us as we begin to redefine the data center of the future and get ready for smarter and more sustainable operations to support greater connectivity and capacity.
In the premier podcast episode, Wendi Runyon, VP of Strategy and Innovation, Electrical Distribution Division, Schneider Electric North America and Adil Attlassy, Chief Technical Officer, Compass Datacenters discuss their ongoing journey to predictive analytics. Until now, analytics have been reactive and underutilized in mission critical environments. Adil says it’s time for a shift.
Lessons from the Airline Industry
Adil reveals why other industries are so far ahead in this area, for instance, the commercial aircraft industry. “Since the 1940s, the fatal accident rate was reduced by 500%.” He credits data as primarily responsible for this incredible result.
Manual, visual inspection was replaced with data driven instrumental condition and statistically based maintenance. Today’s planes, can generate between five to eight terabytes of data per flight.
The goal and ultimate benefit of predictive analytics in data centers is also risk reduction. As such, Adil asserts that equipment should be maintained, not just according to a standard — and fairly random — cycle, but because we have solid analysis to decide when the right time is.
3 Main Steps for Predictive Analytics
Yet, Wendi points out, the data center industry struggles to look at data, let alone create value with it. The two agree that this is mainly because taking a system wide approach is not the current practice. Failure and degradation modes are viewed independently when they should be looked at holistically to gauge how the entire system behaved based on a failure. Then, the right questions can be asked and valuable insights gained.
Adil emphasizes that “since we can’t improve nor analyze what we don’t measure” this process should begin at the fundamental physical telemetry layer to derive and extract as much data as possible. Connectivity between the subcomponents will provide the needed system level view.
At a higher level, he outlines the next three steps for breaking through to predictive analytics as:
- Include more domain expertise in analytic models
- Leverage AI and machine learning more
- Build an asset-based model that can scale
Listen to the Podcast for More Insights
Adil and Wendi’s conversation dives more into how they’re planning to achieve a vision of predictive analytics together. They also talk more about the benefits of predictive analytics, the concerns around cyber security and how long it might take until they reach their ultimate goal. Listen to the Data Center Tech Bytes podcast now.
The post From the Basics to Breakthrough: How to Achieve Predictive Analytics appeared first on Schneider Electric Blog.