Multimedia Checkpoint Protocol with Lower Recovery Overhead

S. Osada and H. Higaki (Japan)



In order to acheive fault-tolerant distributed systems, checkpoint-recovery has been researched and many protocols have been designed. A global checkpoint taken by the protocols have to be consistent. For conventional data communication networks, a global checkpoint is defined to be consistent if there is no inconsistent message in any communication channel. For multimedia communication networks, there are different requirements for time-constrained failure-free execution and large-size message transmissions where lost of a part of the message is acceptable. This paper proposes novel criteria, consistency and duplexity, for global checkpoints in multimedia communication networks.

Important Links:

Go Back