We want to have the ability to run concurrent r/w workload while the cluster is balancing. We should be able to coordinate the start/end of the clients applying their workload and collect the operation metrics afterwards.

Parameters:
1. ratio for the type of operation (regular r/w vs transactional vs retryable)
2. total combined offered load; clients shouldn't apply more than this aggregate amount to the server
3. hosts to run the clients on