Getting to know topology self-discovery, the blast radius, predictability, and correlation

This is where monitoring gets interesting. With normalized and combined MELT data, we can use more advanced data science techniques to uncover what’s happening inside a system based on certain patterns in a dataset.

AI models are becoming a standard feature offered in most APM suites. An AI model is nothing more than specialized software built with artificial neural networks. Those algorithms can be trained to recognize patterns, including anomalies. If what characterizes normal and abnormal behavior is understood by the AI model, it can predict failures before they impact the user by checking whether the behavior is trending in a certain direction.

AI models ingest MELT data to do machine learning. This first set of data is called training data, which tells the AI model what’s good. Then, it creates a hypothetical model of the data to find patterns inside it. The next step...