A queuing server is used for reliable message transport, where one subsystem desires to execute one or more ordered operations asynchronously. Messages are sent to the queue in groups, which may have one or more messages. Messages within a particular group are processed in a predetermined order. Optionally, groups of messages can marked as correlated such that all groups within a particular correlation can be processed in a predetermined order. A message can be stored in a SQL database table until processing of that message is complete. The receiving side of the message system can be scaled across multiple machines and/or across available resources of any given machine. The system can handle “disaster” scenarios on both the sending side (i.e. the sending machine crashes in the middle of sending a group), and the receiving side (i.e., a power failure causes a reboot in at least one of the receiving machines).