A computer-implemented includes a computing system receiving one or more queries. The computing system includes one or more compute nodes that perform computations for determining a response to at least one query. The system stores, in a storage device, domain data that includes at least one of: a dataset, a metric associated with the domain data, a query time, or a usage pattern that is based, in part, on the one or more queries. The method includes the system generating a distribution model based on analysis of the domain data. The distribution model is generated using machine learning logic executed by the system. The method further includes the system using the distribution model to distribute data to the one or more compute nodes. The distributed data is used to determine, within a threshold response time, the response to the at least one query.