Techniques facilitating advising bid values for usage of resource in a resource allocation data analytics framework are provided. In one example, a computer-implemented method can include generating, by a system operatively coupled to a processor, one or more performance models. A performance model of the one or more performance models is associated with respective service classes for a resource of a computing infrastructure associated with a provider network and comprising one or more resources available for allocation. A service class of the respective service classes represents a performance differentiation compared to other ones of the respective service classes. The computer-implemented method can also include determining, by the system, a suggested bid value for the resource. The suggested bid value represents an expected performance metric of the resource as a function of a bid value and is based on the one or more performance models.