Distributed In-Memory Key-Value Store with Raft Consensus

Architect a distributed fault-tolerant key-value database in Go with leader election, log replication, snapshotting, and strict linearizability guarantees.

GoRaftConsensusgRPCDistributed SystemsLevelDB

Fork Starter Repo Live System Demo

Architectural Overview

### Distributed Raft Key-Value Cluster Architecture 1. **Consensus Module:** Executes term voting, leader heartbeats, and append-entries RPCs conforming to the Raft specification. 2. **State Machine Interface:** In-memory high-throughput hash-table persisted by write-ahead log (WAL) segments. 3. **Linearizability & Client Sessions:** Strict read-after-write consistency guaranteeing linearizability for all external clients.

Target Engineering Competencies

Implement leader election and randomized heartbeat timers with exponential backoff

Replicate state machine logs across quorum nodes with commit index safety

Implement state machine snapshotting and log compaction to prevent memory exhaustion

Defeat network partition split-brain anomalies using majority quorum checks

Recommended Technical Prerequisites

Concurrency Primitives in Go (Goroutines, Channels, sync.Mutex)

TCP/gRPC Foundations and Protocol Buffer serialization

CAP Theorem, Quorum arithmetic, and Network Split Invariants

100-Point TA Evaluation Rubric

Passing Benchmark: 80 / 100

EVALUATION CRITERION	DESCRIPTIVE BENCHMARK	POINTS
Leader Election Invariants	Randomized election timeouts, split-vote avoidance, and leader authority transfer.	25 pts
Log Replication & Commit Safety	Strict log matching, majority commit quorum, and uncommitted entry rollback.	30 pts
State Machine Snapshotting & Compaction	Memory snapshot generation, log truncation, and install-snapshot RPC handling.	25 pts
Code Cleanliness & Concurrency Testing	Race condition prevention (`go test -race`) and chaos failure injection resilience.	20 pts

Submit for Rubric Evaluation

Push your implementation to GitHub and submit your repository. Verified Teaching Assistants grade against the 100-point rubric within 48 hours.

L8 Code Review Guarantee

Every submission receives structured comments on architecture, concurrency safety, memory footprint, and test coverage directly from ex-Google & Microsoft faculty engineers.