Examples described herein provide for a memory and at least one processor coupled to the memory. The at least one processor indicates a prediction of a performance goal failure based on performance monitoring of the at least one processor. The performance goal can be based on a service level agreement (SLA). The performance monitoring can be related to core activity or inactivity. A trained machine learning (ML) model can be used to infer performance goal failure based on performance monitoring of the at least one processor. The ML model can be trained using a simulation of traffic to use a compact set of performance monitoring indicators. Mitigation efforts can take place to avoid violation of the SLA.