Distributed application entity groups: analysis and simulation results

Results of analyzing and simulating the distributed application entity group (DAEG) mechanism for application reliability are described. The DAEG mechanism provides fault tolerance at the software level by replicating critical application entities, such as processes. A leader/follower model is used where all inter- and intra-group communication is coordinated by the leader. Our investigation showed that the DAEG mechanism provides fault tolerance under a wide variety of conditions. We determined that the resiliency of the DAEG mechanism imposes a high cost in terms of network traffic under normal circumstances, but not significantly more when failures occur.<
>