Abstract: A computer-implemented method for identifying a cause of a performance anomaly of a computer system executing workloads in different workload groups is disclosed. The method comprises receiving system performance data, separating contention-related data and non-contention related data within the received system management data, feeding a first part of the contention-related data to a first machine-learning system comprising a trained first machine-learning model for predicting first contention instances and related first impact values as output, and feeding a second part of the contention-related data scaled with the first impact values to a second trained machine-learning system comprising a trained second machine-learning model for predicting second contention instances and related second impact values for the different workload groups as output.

Abstract: Aspects of the invention include determining, by a machine learning model, a predicted workload for a system and a current system state of the system, determining an action to be enacted for the system based at least in part on the predicted workload and the current system state, enacting the action for the system, evaluating a state of the system after the action has been enacted, determining a reward for the machine learning model based at least in part on the state of the system after the action has been enacted, and updating the machine learning model based on the reward.

Abstract: Performance predictions in a computing cluster can be provided by sampling and storing historic workload request data of the computing cluster as time-stamped workload values, forecasting an expected total number of workload requests for a defined time interval in the future based on a time-series analysis of the time-stamped workload values, where the time-series analysis detects cyclic and repeating events in the time-stamped workload values. In response to a result of the time-series analysis, training a workload prediction model by using additional data about acyclic events in expected workload requests, where the training applies a statistical regression technique for predicting a future workload demand for the computing cluster, and in response to exceeding a predefined threshold value of the predicted future workload demand, reassigning resources of the computing cluster.

Abstract: Aspects of the invention include determining, by a machine learning model, a predicted workload for a system and a current system state of the system, determining an action to be enacted for the system based at least in part on the predicted workload and the current system state, enacting the action for the system, evaluating a state of the system after the action has been enacted, determining a reward for the machine learning model based at least in part on the state of the system after the action has been enacted, and updating the machine learning model based on the reward.

Abstract: Performance predictions in a computing cluster can be provided by sampling and storing historic workload request data of the computing cluster as time-stamped workload values, forecasting an expected total number of workload requests for a defined time interval in the future based on a time-series analysis of the time-stamped workload values, where the time-series analysis detects cyclic and repeating events in the time-stamped workload values. In response to a result of the time-series analysis, training a workload prediction model by using additional data about acyclic events in expected workload requests, where the training applies a statistical regression technique for predicting a future workload demand for the computing cluster, and in response to exceeding a predefined threshold value of the predicted future workload demand, reassigning resources of the computing cluster.