The major issue we want to address is how to handle omission faults in distributed computing systems. In order to do so, we consider a simple multiclient-multiserver model where it is necessary to solve the reliable atomic broadcast problem. The properties sought are stronger and the fault assumptions are more realistic than those usually considered. In particular, we are interested in investigating the issue of receive omission faults. A fault model is used that encompasses crash and omission faults over processes and communication links. Previous work is briefly overviewed and a solution to the reliable atomic broadcast problem, based on quorums and global sequencers, is sketched out.