Few mainframe shops can afford the luxury of a dedicated performance monitoring team. Those that do, dedicate their time to learn system behaviors and set meaningful thresholds to watch for out of bounds metrics indicating a problem has or is about to occur. This is a very time-consuming effort that requires regular review and maintenance to avoid the constant “that light is always red, ignore it” scenarios. Shops that do not have a dedicated team may set some thresholds on critical workloads or system components, but rarely get the opportunity to consistently review those thresholds; for example, a CPU threshold set three hardware versions ago is now obsolete. What if you could utilize Machine Learning to automate this process? What if your system could learn from your data? Learn what is normal on a Monday morning or a Thursday night?
In this session, come learn how Machine Learning can relieve you of this workload and let you focus on actually tuning the workloads, not just fighting fires.