RAFT_IMPLEMENTATION

Raft Consensus Implementation

Consensus indicates that several servers agree on the same information, which is required for designing fault-tolerant distributed systems. Consensus algorithms enable a group of computers to function as a coherent entity that can withstand the failures of some of its members. RAFT protocol helps in achieving consensus.

In this project I implemented the two major task involved in RAFT algorithms

• Leader Election
• Safe Log Replication

Leader Election

As per RAFT algorithm each node in a cluster can exist in three states, namely Leader, Candidate and Follower. The below flow diagram explains the transition of nodes into different states.

For implementing succesful Leader Election, I incorporated below mentioned mechanism on each node

• Heartbeats
• ElectionTimeout
• AppendRPC
• RquestRPC

Block diagram showing the implementation details

Please check this file to see more information on the Leader Election Implementation

Safe Log Replication

Implemented Safe Log Replication along with modified Leader Election algorithm to ensure only a candidate with complete log must get elected as leader. The below flow diagram explains the log replication process.

Implementing of safe log replication involves implementation of following mechanism

• Leader Completeness - Candidates with incomplete logs must not be elected as leader.
• Log replications to all the followers by the leader.
• Log ordering - Ensure the integrity of log.
• Log inconsistencies - Managing Follower’s log with conflicting log entries or incomplete log entries.

Please check this file to see more information on the Safe Log Replication Implementation

Tools and Technologies

• Java 8
• Java Spring
• Multithreading in Java
• UDP Datagram Packets - Used UDP packets for sending heartbeats, vote requests and other messages
• Docker - Built 5 container to simulate algorithm

How to use this repository

• Please check Readme.pdf here

Testing and Validation

• please check validation.pdf here

Demo Video

Please check the demo video of the implementation here

References

• https://docs.docker.com/storage/volumes/
• https://docs.docker.com/engine/reference/builder/
• https://docs.docker.com/compose/
• https://docs.docker.com/compose/networking/
• https://raft.github.io/#implementations
• https://howtodoinjava.com/java/multi-threading/java-thread-pool-executor-example/
• http://thushw.blogspot.com/2011/06/asynchronous-udp-server-using-java-nio.html