Distributed Key-Value Store

Overview

This project is a distributed key-value storage system implemented in Java, designed for high throughput, low latency, and strong consistency. It uses Multi-Raft consensus, sharding, and a hybrid storage engine to support scalable and reliable data access.

Features

• Multi-Raft consensus for replication and consistency
• Consistent hashing with virtual nodes for load balancing
• Dynamic shard rebalancing
• Hybrid storage: RocksDB (LSM-tree) + in-memory B+ Tree
• MVCC snapshot isolation
• Read optimizations (ReadIndex, FollowerRead)
• Batched asynchronous writes

Performance

• ~50K QPS throughput
• <10ms P99 latency
• 99.9% availability

My Contributions

• Designed and implemented the system architecture
• Built consistent hashing and shard rebalancing
• Implemented storage layer with MVCC
• Optimized read/write paths for performance

Technologies

• Java
• Distributed systems (replication, consensus, sharding)
• RocksDB