Abstrait

Fused Data Structure for Tolerating Faults in Distributed System

Jeva Kumar M, Golda Jeyasheeli P

In Distributed systems, servers are prone to crash faults in which the data structures (queue, stack, etc) may crash, leading to a total loss in state. Hence it is necessary to tolerate crash faults in distributed system.Replication is the prevalent solution to tolerate crash faults .In replication, entire copy of the original data is taken and stored. Every update to original data reflects changes in the replicated data. Replication is used to ensure consistency, improve reliability, fault-tolerance and accessibility.To tolerate f crash faults among n distinct data structures, replication requires f replicas of each data structure, resulting in nf additional backups.Maintaining nf additional backups for n distinct data structures requires more space. In this project, fused data structure is used for backups which can tolerate f crash faults using just f additional fused backups.