Machine Learning (ML) is all around us. According to a new Constellation Research Survey, 70% of organizations are using it in some form. The rise of Machine Learning is supposed to transform the nature of work and collaboration between humans and machines. In other words, leave the mundane to the machines and focus your time on solving business problems.
Sounds logical. But in practice, it's temptingly counter-intuitive. Perhaps because it's hard to get over what we have practiced for decades. I mean, we know which metrics to track and even the precise thresholds to set alarms for. Right? Well, as it turns out in making such decisive calls, we may have contributed to massive numbers of false alerts. A 2018 report by DevOps.com states that one in five IT organizations receive over 1000 operations alerts each month. This proves the hard, cold fact that relying on your intuition or previous experience may not be the best way to deal with IT Operations and security issues.
Closer to home, think about a lift-and-shift scenario. Organizations are actively involved in moving their applications and databases to the cloud. Their primary motivation is to get out of the data center management business but as they do so, performance monitoring and security are primary concerns: will my performance in the cloud be as expected? how does it compare to my data center performance? how do I do capacity planning with resources in the cloud? how do I secure my application against increasing threats? As I have written previously, ML helps with each one: anomaly detection for performance and security, clustering to understand outliers, top (N) SQL anomalies for performance variability, forecasting based on observed dynamic consumption patterns, understanding configuration drift, user behavior analysis and finally auto-remediation workflow actions are among many others. ML simplifies what is inherently very difficult to do manually.
The same general principle applies broadly to much of systems and security monitoring. So, and rightfully, argues an Oracle Management Cloud Specialized Partner, Erik Benner of Mythics. In a recently published Forbes OracleVoice article, Erik forcefully argues for curbing such instinct and letting ML do its job without interference. Erik busts some common Systems management myths, arguing that Machine Learning can be very helpful in unexpected ways.
Want to get a quick hands-on experience with Oracle Management Cloud? Use our free EasyStart Kit and get started in under and hour!