You are here

You are here

IT Ops analytics: How to keep up with continuous delivery

John P. Mello Jr. Freelance writer

Over the past five years, tools for analytics have made inroads with IT operations teams, including integration into IT Ops tools themselves. “People started talking about IT operations analytics a few years ago, but it didn’t take off,” explained Nancy Gohring, a senior analyst with 451 Research.

The early technology really wasn’t there, Gohring said. “The initial products didn’t scale very well, and they really didn’t have great analytics tools. There’s a lot more sophisticated technology now, both in analytics and machine learning.”

One driver behind the move toward IT operations analytics is that systems are becoming more complex and the data they’re creating is exploding.

“You used to be able to almost visually look at the raw data and see what the problem was. But now there’s so much data coming in, a human can’t analyze it, so you need machine learning and analytics to churn through all that data and make sense of it.”
Nancy Gohring

Keep up with the DevOps and CD crowd

Another big reason organizations are embracing IT operations analytics is to keep up with the speed of DevOps and continuous delivery (CD). The faster that problems can be found and resolved, the more time that IT Ops has to deploy and maintain all those releases.

Business alignment is also a driver behind IT operations analytics: By training your monitoring tools on key performance indicators, and knowing quickly what may be slowing things down, IT can help the business keep humming along. 

“We have all these monitoring tools—application monitoring tools, server monitoring tools, log management tools, data base monitoring tools—each with their data in silos,” Gohring explained. “When a problem in an application occurs, all those tools can start throwing up alerts. That makes it difficult to find the root cause of a problem.”

“Machine learning and analytics can break down those silos so the data from the monitoring tools can be correlated and the problem fixed.” 

Get up to speed fast on IT Ops analytics

This week, TechBeacon Learn launches a new track designed to get you up to speed with IT operations analytics, so your team can be better prepared to resolve issues quickly, save time and money, and run at the speed of DevOps.

For starters: four modes of analytics. You’ll start by reviewing the four basic modes of analytics, along with use cases that show how each works at a practical level:

  • Descriptive analytics presents data in the form of dashboards, quick reports, data consolidation, search/data mining, scorecards, and key performance indicators so you can more precisely understand the nature and extent of operational problems.
  • Diagnostic analytics helps pinpoint the cause of those problems. It includes automated root-cause analysis, intelligent notification, collaborative investigation, and real-time analytics.
  • Predictive analytics uses machine learning to identify IT operations data associated with a past problem. It uses those insights to suggest when another similar problem, say a failure, is likely.
  • Prescriptive analytics combines the three modes described above and delivers specific suggestions to prevent or mitigate specific outages or cost spikes.

Mo Patel, practice director for artificial intelligence and deep learning at Think Big Analytics, said:

“Step one is to use analytics to see what has happened, but then organizations need to get beyond that and into the prescriptive and predictive. We should be able to see things coming.”

Analytics maturity

Subject matter experts explain how these modes of analytics are additive—i.e., they work together as teams grow in sophistication and deliver real value to the business. Real-life scenarios provide the context for you to understand what to monitor and how to move from a reactionary to a proactive stance in day-to-day IT operations.

“Being able to collect data and establish metrics from everything to how long does it take an app to open to how long does it take to resolve a ticket enables you to spend money in the right places.”
—Mo Patel

Machine learning and more

How do you teach a machine? With algorithms and data, lots of data, whose characteristics you describe as valuable, or not, for your specific needs. Over time, the machine “knows” what you’re looking for and can spot threat indicators or other anomalies that you can explore with root-cause analysis.

Upcoming units in this track will cover anomaly detection, behavioral analysis, hybrid cloud management techniques, compliance, alerting, and other topics through short, digestible articles and tutorials.

Keep learning

Read more articles about: Enterprise ITIT Ops