Redis Inference Benchmark

Introduction

This repository benchmarks and compares three popular message brokers:

• RabbitMQ - Robust message queue with advanced routing
• Redis - In-memory data store with pub/sub capabilities
• Kafka - Distributed event streaming platform

The goal is to evaluate performance metrics and identify the best broker for your network requirements.

Quick Start

Prerequisites

Before running the benchmark, ensure you have set up:RabbitMQ , Redis and Kafka .

Then install Python dependencies:

pip install -r requirements.txt

Running the Benchmark

Start the Sender:

python main.py --role sender

Start the Receiver (in another terminal):

python main.py --role receiver

The sender and receiver will communicate through the configured broker, measuring throughput and latency.

Configuration

Edit cfg/config.yaml to customize the benchmark:

# Broker Configuration
rabbit:
  address: "localhost"
  username: "guest"
  password: "guest"
  virtual-host: "/"

redis:
  host: "localhost"
  port: 6379
  db: 0

# Benchmark Settings
num_rounds: 100          # Number of messages to send
message_size: 15         # Message size in MB (max 16 MB)

Project Structure

├── main.py              # Entry point (RabbitMQ, Redis, Kafka runners)
├── client.py            # Client utilities
├── src/
│   ├── Utils.py        # Configuration loader
│   ├── rabbitmq/       # RabbitMQ sender/receiver
│   ├── redis/          # Redis sender/receiver
│   └── kafka/          # Kafka sender/receiver
├── cfg/
│   └── config.yaml     # Broker configuration
│ 
└── requirements.txt    # Python dependencies

Metrics

The benchmark measures:

• Message throughput (messages/sec)
• Latency (milliseconds)
• Resource utilization

Notes

• Maximum message size: 15 MB
• Default rounds: 100 rounds
• Ensure all brokers are running before starting the benchmark