Handling Failures
Clients can’t directly detect what is really going on--both good and bad
- Can have replicated server objects
- Can transparently handle failures by spawning new server objects
- Client application has to have other ways to react to failures that are not contained by the ORB