This project implements a simple distributed day trading service with the core goal of using the Raft consensus algorithm to ensure consistency across a distributed system.

Kotlin is used for the implementation of the service, and gRPC is used for the communication between the distributed servers, and between the client and the servers.

At a high-level, the system will consist of a set of servers that will be responsible for maintaining a replicated log of client requests. The servers will use the Raft consensus algorithm to ensure that the logs are consistent across all servers. The leader node is responsible for communicating with the client and replicating the log across all other nodes.

---
title: Raft State Machine
---
stateDiagram-v2
    [*] --> Follower: Startup or restart
    Follower --> Candidate: Suspects leader failure
    Candidate --> Leader: Receives quorum
    Candidate --> Candidate: Election times out
    Candidate --> Follower: Discovers new term
    Leader --> Follower: Discovers new term

The implementation of the Raft state machine can be found in RaftStateMachine.kt, with the implementation of each specific state, event, and action being in Node.kt.

1. Install docker. You can find the instructions for your OS here: https://docs.docker.com/get-docker/
2. Build the repository (generates the jar file) with ./gradlew build
3. Create a Docker image with the jar file with the following command: docker build -t raft-trading-server:latest .
4. Create a Docker image for the frontend with the following command: docker build -f Dockerfile-frontend -t raft-trading-frontend:latest .
5. Run the simulated environment with 3 containers representing 3 separate raft nodes: docker compose up -d
   • Note that the -d flag is use for running in daemon thread. You definitely WANT to do this, especially on compose, since if you don't include it, the container(s) will be tied to the terminal session in which it was called, and will stop if you hit ^C or close the window.
   • The frontend runs on http:https://localhost:80
     • You can use any login parameters, for example:
       1. Fill in email and password such as cs416 and password raft (there's no security on this so don't put anything sensitive)
       2. Click on Register to create account - nothing happens and that's ok
       3. Click on Login to close the modal
     • You can then deposit any amount into your balance, and then purchase any stock such GME for 5 shares.
     • Selling shares will also remove those and add balance to account

Docker Desktop and Intellij IDEA are great applications for observing the logs of each Docker container (representing a single Raft node) and the progress of the Raft algorithm.

Postman can be used to send client requests to the leader node over gRPC. To do this, you will need to import the trade.proto file so you can access the gRPC methods. Note that we use port forwarding to access the gRPC server running on the Docker container from you local device. The port forwarding is defined in the docker-compose.yml file.

The number of nodes (Docker containers) used by the Raft can easily be changed by adding additional services to the docker-compose.yml file and updating the config.json with the new node information.

• Support dynamic membership changes (Section 6 of Raft paper) (i.e. adding/removing nodes from the cluster)

• Implement snapshotting and sharing of snapshots between nodes (Section 7 of Raft paper)

‼️ Note that this repository is for educational purpose and should not be used in production!

• In Search of an Understandable Consensus Algorithm (Extended Version)
• Raft Kotlin