Uso de Broadcast na Sincronização de Checkpoints em Protocolos Minimais

Tiemi C. SakataIslene C. GarciaLuiz E. Buzato

In a synchronous checkpoiting, the application can be easily recovered in case of failures because the processes may rollback to their last checkpoint on stable storage. This article examine the minimal protocols, in which just a minimal number of processes take checkpoints to construct a consistent global checkpoint. Cao and Singhal propose a new approach to develop a minimal protocol. This approach uses a broadcast to block all processes and centralizes to a unique process the task of determine which processes should take checkpoints during the consistent global checkpoint construction. This article contains a prove the protocol proposed by Cao and Singhal is not minimal and we propose a correction to change the protocol and to guarantee the minimality.

Caso o link acima esteja inválido, faça uma busca pelo texto completo na Web: Buscar na Web

Biblioteca Digital Brasileira de Computação - Contato:
     Mantida por: