Redis vs DynamoDB Benchmarking

A performance comparison tool for Redis and DynamoDB, designed to benchmark read and write operations with vector embeddings to help you choose the right database for your use case.

What We're Benchmarking

This project compares:

• Redis - High-performance in-memory data structure store
• DynamoDB - AWS managed NoSQL database with automatic scaling

Benchmark Operations:

• ✅ Write Performance - Insert operations throughput and latency
• ✅ Read Performance - Query operations throughput and latency

Tutorial & Data Modeling Guide

📚 Redis Embeddings Data Modeling Tutorial - Interactive Jupyter notebook covering:

• Redis data structures and APIs for vector embeddings
• Best practices for data modeling patterns
• Hands-on examples with real embedding workflows
• Performance optimization techniques

Recommended reading before running benchmarks to understand the underlying data patterns being tested.

Quick Start

Prerequisites

• AWS EC2 instance (Amazon Linux 2023 recommended)
• Python 3.11+
• Redis server (for Redis benchmarks)
• AWS credentials configured (for DynamoDB benchmarks)

Setup & Installation

1. System Dependencies (AWS EC2 - Run Once)

# Update system and install essential tools
sudo dnf update -y
sudo dnf install -y git curl wget gcc gcc-c++ make libffi-devel openssl-devel python3-pip

# Verify Python 3 is available
python3 --version

2. Python Environment Setup

# Clone the repository
git clone https://github.com/redis-applied-ai/dynamo-vs-redis
cd dynamo-vs-redis

# Install uv (fast Python package manager)
python3 -m pip install --upgrade pip pipx --user
export PATH="$HOME/.local/bin:$PATH"
pipx install uv

# Create virtual environment and install dependencies
uv sync

3. Run Benchmarks

# Basic Redis benchmark (runs both write and read tests)
uv run python benchmark.py --db redis

# Basic DynamoDB benchmark (runs both write and read tests)
uv run python benchmark.py --db dynamo

# Advanced usage with custom settings
uv run python benchmark.py --db redis --proc 4 --ops 1000
uv run python benchmark.py --db dynamo --proc 16 --ops 5000

Benchmark Results

Here are benchmark results comparing Redis vs DynamoDB performance with default settings (--ops 10000 on 8 processes):

Redis Cloud Results

Amazon DynamoDB Results

Key Findings: With no tuning on either database — besides standard VPC peering for both — these are the out-of-the-box results for simple read/write tasks involving vector embeddings (cast to bytes).

Configuration

Environment Variables

For Redis:

• REDIS_HOST - Redis server hostname (required)
• REDIS_PASSWORD - Redis password (required)
• REDIS_PORT - Redis port (default: 6379)

For DynamoDB:

• DDB_TABLE - DynamoDB table name (required)
• AWS_REGION - AWS region (default: us-east-1)

Command-Line Options

The benchmark script accepts these arguments:

• --db {redis,dynamo} - Database to benchmark (required)
• --proc N - Number of parallel processes (default: 8)
• --ops N - Total number of operations per test (default: 2000)

Note: Each benchmark run executes both write and read tests automatically, providing comprehensive performance metrics including QPS, P50, P95, and P99 latencies.

Project Structure

dynamo-vs-redis/
├── benchmark.py                               # Main benchmark script
├── redis_embeddings_data_modeling.ipynb     # Tutorial notebook  
├── pyproject.toml                            # Python dependencies
├── REDIS.png                                 # Redis benchmark results
├── DYNAMODB.png                              # DynamoDB benchmark results
└── README.md                                 # This file

Use Cases

This benchmarking tool is ideal for:

• 🚀 AI/ML Applications - Vector similarity search and embedding storage
• 📊 Performance Testing - Database selection for high-throughput workloads
• 🔍 Cost Analysis - Comparing operational costs between Redis and DynamoDB
• 🏗️ Architecture Planning - Making informed decisions for production deployments