By Jeff Henry, VP Product Management, CA Technologies
In my previous blog in this series, I introduced the topics of AI, machine learning and predictive analytics. And I shared my belief that it’s no longer necessary to engage a data scientist to enjoy the benefits of machine learning in IT operations. This time, I’m going to take a closer look at some of the specific ways machine learning and predictive analytics help enterprise IT departments overcome some of their biggest challenges.
Let’s start with a core platform: the mainframe. Mainframes remain mission-essential in today’s app economy, but the skills and know-how to manage them are harder and harder to find. And yet, 55% of apps touch a mainframe, including those designed to process credit card and airline transactions, and more. Mainframe uptime is absolutely fundamental to business performance – impacting everything from user experiences on mobile apps to customer satisfaction to your ability to transact with customers.
What do these demands mean for IT operations teams? They mean pressure to mitigate risks so the mainframe environment stays healthy and always-on. And they create a critical need to prevent issues from occurring, and resolve them proactively if and when they happen.
Reactive or proactive?
Now think about your current mainframe monitoring and management. Is the emphasis on how fast you can react to problems once they’ve happened (MTTR), rather than anticipating issues before they bite? You’re not alone. As mainframe systems and applications become ever more complex and system knowledge decreases over time, the time taken to triage problems and repair failures often increases. The challenge of retaining mainframe experts with the skills to deal with these failures is growing as they reach retirement age. And as if that weren’t enough, the penalties for failing to meet business SLAs are increasingly high.
Predictive analytics in action
Machine learning enables you to overcome these challenges with intelligent automation. Rather than wait until a pre-defined threshold is reached before sending an alert, predictive analytics algorithms learn the patterns in your mainframe monitoring data. Not only do you get alerted earlier to anomalies sooner – before they turn into outages. You get fewer false positives that lead to the “sea of red” that’s all too common in mainframe monitoring environments. And with fewer alerts to attend to, monitoring teams can focus their expertise where it matters, without the anxiety of worrying they’ve missed a critical red flag.
Best of all, these capabilities are increasingly available in your infrastructure and applications monitoring platforms. Next time, my colleague Kartik Thangavelu will look at how you can implement predictive analytics in your own environment – reducing risk, improving performance and helping you sustain SLA compliance.