I am new in ML space and working on the task to build a model that would predict performance degradation of integration bus in real time or "almost real" time. The idea is fairly straightforward, I have logs from the production and filtered required information in pandas dataframe: "Provider\consumer system" (object), "Request Time" (dataframe64), "Response Time" (dataframe64), and "Time Performance" (float). The last column is the difference between the response and request times in ms, it is our target column to predict performance degradation. So, I have a list of providers and consumers sending messages through the integration system. I am thinking about a model that will predict that if some provider\consumer breaches the threshold for some time it would lead to an increase in performance time and will warn about this. So, we will get a warning in seconds\minutes after potential degradation starts. When I researched similar problems, I found out that it may fall under the predictive maintenance category, but not much more about this area of ML. Would appreciate any thoughts about the approach or model that may help.