🚀 InkStore — High-Performance In-Memory Key-Value Store

InkStore is a high-performance, multi-threaded in-memory key-value store written in C++, designed to model real-world caching systems like Redis.

It acts as a caching layer to reduce backend latency and database load by serving frequently accessed data directly from memory.

---

🧠 Problem Statement

Modern backend systems face:

• High latency due to repeated database queries
• Performance degradation under concurrent load
• Inefficient handling of frequently accessed data

💡 Solution

InkStore introduces an in-memory caching layer:

Client → InkStore (cache) ⚡ → Database (fallback)

This significantly improves response time and reduces database pressure.

---

⚙️ Features

🔹 Core

• In-memory key-value storage (O(1) average operations)
• TCP socket-based server
• Multi-threaded client handling

🔹 Advanced

• Write-Ahead Logging (WAL) for durability
• TTL (Time-To-Live) for automatic key expiration
• LRU eviction policy for memory management
• Thread-safe operations using mutex synchronization

🔹 Performance

• Custom multi-threaded benchmarking client
• Throughput & latency measurement
• Concurrency stress testing

---

🏗️ Architecture

Client (ncat / custom client)
        ↓
TCP Socket
        ↓
InkStore Server (C++)
        ↓
In-Memory Store + WAL

---

🧪 Benchmark Results

Threads	Throughput (req/sec)	Latency (ms)
1	~7700	0.12
12	~7300	0.13
50	~4300	0.23
80	~3000	0.33
100	~3100	0.31
150	~3170	0.31
200	~1290	0.77

---

📊 Performance Visualization

🔹 Throughput vs Threads

---

🔹 Latency vs Threads

---

🧠 Performance Analysis

• Best performance observed at low concurrency (1–10 threads)

• Throughput decreases as thread count increases due to:

  • Mutex lock contention
  • Context switching overhead
  • Socket connection overhead

• Latency increases significantly under high concurrency

• Demonstrates real-world limitations of shared locks in multi-threaded systems

---

🛠️ Tech Stack

• Language: C++ (C++17)
• Networking: POSIX Sockets
• Concurrency: std::thread, mutex
• Persistence: Write-Ahead Logging (WAL)
• Data Structures: HashMap + Doubly Linked List (LRU)

---

▶️ Getting Started

🔧 Build

make

▶️ Run Server

./inkstore

⚡ Run Benchmark

make benchmark
./benchmark_app

---

🧪 Example Usage

Using ncat:

ncat 127.0.0.1 9090

SET name abhijeet
GET name
GETALL
DEL name

---

🚀 Real-World Applications

• API response caching
• Session storage
• Rate limiting systems
• High-performance backend services

---

🧠 Key Learnings

• Concurrency control and synchronization
• Performance vs scalability trade-offs
• Memory-efficient caching strategies
• Importance of benchmarking and profiling

---

⭐ Conclusion

InkStore demonstrates core backend infrastructure concepts including caching, concurrency, persistence, and performance analysis — closely mirroring real-world system design patterns used in high-performance services.

---