VKernel's Alex Rosemblat sits down with VP of Product Management Eric Jackson to discuss the proper employment of self-learning analytics.
- Self-learning analytics are very helpful around certain metrics (utilization) and assuming that you are baselining against a stable, well-running environment
- Self-learning is irrelevant for certain other metrics that measure contention or larger problems
- Other techniques are needed in addition, for example, to capture changing trends that may not be immediately problematic, but imply that problems may be coming sooner than expected
- In the end, any alert that something abnormal is happening needs to be interpreted, diagnosed and a remediation figured out. Self-learning provides an important source of alerting, but the rest remains.