Skip to content

README.md

Latest commit

 

History

History
188 lines (128 loc) · 5.56 KB

README.md

File metadata and controls

188 lines (128 loc) · 5.56 KB

RedisSMQ Benchmarks

Pre-release (next) Code Coverage (next)

💡 Note: You are viewing the next branch with upcoming features. For stable releases, check the master branch.

A comprehensive benchmarking suite for RedisSMQ message queue system, measuring throughput, latency, and end-to-end performance under various workloads.

📊 Overview

This benchmark suite provides tools to measure RedisSMQ performance across different scenarios:

  1. Producer Throughput - Measure message production rates
  2. Consumer Throughput - Measure message consumption rates
  3. End-to-End Throughput - Measure complete system performance with concurrent producers and consumers

Note: These benchmarks are for performance testing and should be run in a controlled environment. Always test with production-like data and conditions.

🚀 Quick Start

Prerequisites

  • Node.js 20+
  • Redis server running (default: localhost:6379)

Installation

npm install redis-smq@next redis-smq-common@next redis-smq-benchmarks@next --save

⚙️ Configuration

Benchmarks can be configured using environment variables:

Variable Description Default
BENCH_MESSAGES Total messages to process 100000
BENCH_PRODUCERS Number of producer worker threads 0
BENCH_CONSUMERS Number of consumer worker threads 0
BENCH_SHOW_PROGRESS Benchmark progress reporting 0
REDIS_HOST Redis server host localhost
REDIS_PORT Redis server port 6379
REDIS_DB Redis server database 0

Example Configuration

# Run with custom configuration
REDIS_HOST=127.0.0.1 \
BENCH_PRODUCERS=10 \
BENCH_CONSUMERS=5 \
npx redis-smq-benchmarks

📈 Benchmark Types

1. Producer Throughput Benchmark

Measures how quickly producers can enqueue messages.

Features:

  • Even distribution of messages across producers
  • Real-time progress reporting
  • Individual producer statistics
  • Aggregate throughput calculation

Usage:

BENCH_MESSAGES=50000 BENCH_PRODUCERS=4 npx redis-smq-benchmarks

2. Consumer Throughput Benchmark

Measures how quickly consumers can process messages.

Features:

  • Pre-fills queue with messages
  • Even distribution across consumers
  • Individual consumer statistics
  • Complete consumption tracking

Usage:

BENCH_MESSAGES=50000 BENCH_CONSUMERS=4 npx redis-smq-benchmarks

3. End-to-End Throughput Benchmark

Measures complete system performance with concurrent producers and consumers.

Features:

  • Concurrent producer and consumer execution
  • Production vs consumption phase tracking
  • System backlog monitoring
  • Complete throughput analysis

Usage:

BENCH_MESSAGES=100000 BENCH_PRODUCERS=8 BENCH_CONSUMERS=8 npx redis-smq-benchmarks

📊 Output Metrics

Each benchmark provides detailed metrics:

Individual Worker Metrics

  • Messages processed per worker
  • Time taken per worker
  • Throughput per worker (messages/second)

Aggregate Metrics

  • Total messages processed
  • Total time elapsed
  • Overall throughput
  • Average time per worker

E2E-Specific Metrics

  • Production throughput
  • Consumption throughput
  • System backlog (produced vs consumed)
  • End-to-end latency

📊 Sample Output

Starting end-to-end throughput benchmark...
Queue: benchmarking/queue-1767808640925 | Messages: 100000 | Producers: 10 | Consumers: 5
Setting up producers and consumers...

========== E2E BENCHMARK COMPLETE ==========
Production Phase:
  Total produced: 100000
  Production time: 3.35s
  Production throughput: 29824 msg/s

Consumption Phase:
  Total consumed: 100000
  Consumption time: 6.37s
  Consumption throughput: 15689 msg/s

End-to-End:
  Total time: 6.37s
  Overall throughput: 15689 msg/s
  System backlog: 0 messages
  Status: All messages processed successfully ✓
============================================

🏗️ Architecture

Design Principles

  1. Worker-Based Architecture: Each producer/consumer runs in its own worker thread
  2. Even Distribution: Messages are evenly distributed across all workers
  3. Real-time Monitoring: Progress updates during execution
  4. Graceful Shutdown: Clean termination of all resources
  5. Extensible: Easy to add new benchmark types

Performance Tuning Tips

  1. Worker Count: Increase worker count to leverage multi-core CPUs
  2. Redis Configuration: Ensure Redis has sufficient memory and connection limits
  3. Network: Local Redis provides the best performance. Network latency impacts throughput

📊 Interpreting Results

Key Metrics to Monitor

  1. Throughput Plateau: Point where adding more workers doesn't increase throughput
  2. CPU Utilization: Monitor CPU usage during benchmarks
  3. Memory Usage: Watch for memory leaks in long-running tests

License

RedisSMQ Benchmarks is released under the MIT License.