Benchmark MySQL

Description

This application benchmarks MySQL query performance using DuckDB and Parquet files with Golang.

Features:

1. Query performance in MySQL:

   • Executes a query directly on the MySQL database and measures execution time.
   • Generates a Parquet file with the query results.

2. Query performance in Parquet files using DuckDB:

   • Reads the Parquet file generated from MySQL data and executes SQL queries on it using DuckDB.

---

Installation

Prerequisites

1. Install Go (v1.23.3 or higher): Download here.
2. Install DuckDB (optional for external verification): Download here.
3. Have access to a MySQL database.

Steps

1. Clone the repository:

2. Install the dependencies:

   go mod tidy

Run

go run main.go

Build

If you want to build the application, put the .env file in the same directory as the binary file.

For Windows:

GOOS=windows GOARCH=amd64 go build -o out/benchmark-mysql.exe main.go

For Linux:

GOOS=linux GOARCH=amd64 go build -o out/benchmark-mysql main.go

For MacOS:

GOOS=darwin GOARCH=amd64 go build -o out/benchmark-mysql main.go

Results

Duckdb is five times faster than MySQL for the same query.

Parquet files are a good compression format for data storage and can be used to improve query performance. 116235 rows where stored in a 2.881MB file.