Latecomer and crash recovery support in fault-tolerant groupware was discussed. A simulated multicast protocol based on TCP was implemented using an axisymmetric total-ordering algorithm. System performance scales well when the number of concurrent joins is relatively small. The solution offers a dynamis trade-off between performance scalability and cost constraint.
|Original language||English (US)|
|Number of pages||14|
|Journal||IEEE Distributed Systems Online|
|State||Published - 2001|
All Science Journal Classification (ASJC) codes
- Hardware and Architecture
- Computer Networks and Communications