Simple RAFT Implementation in Java

This repository features a simple RAFT implementation designed specifically for educational purposes, aimed at helping you grasp the intricacies of the consensus algorithm described in the seminal RAFT paper available at RAFT Paper. For a practical demonstration, you may explore the visual insights provided by The Secret Lives of Data.

Quick Start

To get started, simply run the main method found in the Test class.

Key Features

• Leader Election: The system successfully elects a leader among nodes, ensuring a cohesive decision-making process.
• Log Replication: It effectively replicates logs across nodes to maintain consistency and durability.
• Planned Expansion: Features like Membership Change and Log Compaction are acknowledged for future development, currently left unimplemented as they are not essential to the primary educational objective.

Important Notes:

• This project is strictly for educational purposes and the features are not optimized for production deployments.
• Membership change and log compaction features are currently on hold as the focus remains on successful leader election and log replication.

Future Enhancements and TODOs

The following features have been already implemented:

• Network Emulation: Emulates various network behaviors including joining/leaving networks and handling message transport through an asynchronous messaging queue, including jamming simulations for testing resilience.
• Timer Implementation: Offers scheduled tasks, timeout adjustments, and listen/notify capabilities.
• RAFT RPC: Supports critical RPC calls including AppendEntries and RequestVote with successful test validations.
• Client Commands: Integration of client-side operations such as deposit and withdraw functionalities, with adequate testing.
• Finite State Machine (FSM): A robust transition table and state change listener to manage the RAFT states efficiently.

For further exploration of the inner workings and features of the project, dive into the codebase to learn about the state of leader, follower, candidate, and their respective functionalities.

Join us in understanding RAFT's consensus algorithm deeper and contribute if you wish!