A data warehouse contains data originating from autonomous sources. Various maintenance policies have been suggested which specify when and how changes to a source should be propagated to the data warehouse. Engström et al.(HS-IDA-TR-00-001) present a cost-based model which makes it possible to compare and select policies based on quality of service as well as system properties. This paper presents a simulation environment for benchmarking maintenance policies. The main aim is to compare benchmark results with predictions from the cost-model. We report results from a set of experiments which all have a close correspondence with the cost-model predictions. The process of developing the simulation environment and conducting experiments has, in addition, given us valuable insights into the maintenance problem, which are reported in the paper.