Currently, there isn't a friendly way to get an up-to-date state of a cluster (single server, replica set, shards). We need to provide a strongly-typed method that will invoke the appropriate server commands and aggregate the results back in an easy to use object model.