PrefixOS v2.0.0 — Multi-Raft Clustered KV Cache Engine

🚀 PrefixOS v2.0.0 — Multi-Raft Clustered KV Cache Engine

This is the first stable public release of PrefixOS, a high-performance distributed prefix-aware KV cache engine designed for modern LLM inference systems.

✨ Highlights

• 🧠 Prefix-aware 64-way sharded Radix Tree
• ⚡ Zero-allocation slab memory allocator
• 💾 Durable Write-Ahead Log (WAL) with CRC32 verification
• 📸 Copy-on-Write snapshot engine
• 🔄 Canonical Raft consensus implementation
• 🌐 Multi-Raft clustered architecture
• 🛰️ Consistent Hash Ring with virtual nodes
• 📡 gRPC API + REST API
• 📦 Typed Go SDK
• 📈 Prometheus metrics
• 🧪 Comprehensive unit tests, race detection, benchmarks, and CI

---

Architecture

Client
   │
   ▼
Stateless Router
   │
Consistent Hash Ring
   │
 ├── Shard A → Raft Group → Storage Engine
 ├── Shard B → Raft Group → Storage Engine
 └── Shard C → Raft Group → Storage Engine

---

Core Components

• Zero-allocation Slab Memory Manager
• 64-way Sharded Prefix Radix Tree
• Pluggable LRU / SLRU Cache Eviction
• WAL + Snapshot Persistence
• Canonical Raft Replicated State Machine
• Multi-Raft Sharding
• ReadIndex & PreVote extensions
• Go SDK
• gRPC & REST APIs

---

Engineering

• Go
• gRPC
• Protocol Buffers
• Prometheus
• Multi-Raft Consensus
• CRC32 WAL
• Consistent Hashing

---

Quality

• ✅ Unit Tests
• ✅ Race Detection
• ✅ GolangCI-Lint
• ✅ Benchmark Regression
• ✅ Release Build
• ✅ GitHub Actions CI

---

Release Artifacts

Included in this release:

• prefixos binary
• prefixos.sha256
• PrefixOS.zip

---

Thank you to everyone who reviewed, tested, and provided architectural feedback throughout the development of PrefixOS.