Fault Tolerance in Distributed Systems
Overview
What is a Distributed System
Fault Tolerance
Advantages of Fault Tolerance
Byzantine failure
Handling Byzantine Failures
Handling Byzantine Failures (cont.)
Quorum Mechanism
Different phases in Byzantine fault tolerance
Why Commit Phase is needed?
High Availability in Peer-to-Peer Systems
Erasure coding (EC)
Introduction
Related Works
Paper’s Contribution
MODEL AND PROTOCOL
Preliminaries
Protocol Idea
Contd..
Results
Discussion
Current Research Focus
Application characteristics:
Replica Placement:
Future Research
References

Application characteristics:
Two key properties of peer-to-peer applications that impact the use of replication are object sizes and timeliness of object download.
First, larger objects take longer to replicate and naturally motivate the use of block-level replication. However, when conventional blocking is used for large objects, the reliability of the system depends on large number of hosts being available at the same time.
As a result, the availability of an object is inversely related to its size: the larger the object, the worse the availability.
Second, the relationship between when data is requested and the time at which it must be delivered.