#database #development #golang #reading-list #sql

🔗 Go database driver overhead on insert-heavy workloads

The most popular SQLite and PostgreSQL database drivers in Go are (roughly) 20-76% slower than alternative Go drivers on insert-heavy benchmarks of mine. So if you are bulk-inserting data with Go (and potentially also bulk-retrieving data with Go), you may want to consider the driver carefully. And you may want to consider avoiding database/sql.

Some driver authors have noted and benchmarked issues with database/sql.

So it may be the case that database/sql is responsible for some of this overhead. And indeed the variations between drivers in this post will be demonstrated by using database/sql and avoiding it. This post won't specifically prove that the variation is due to the database/sql interface. But that doesn't change the premise.

Not covered in this post but something to consider: JetBrains has suggested that other frontends like sqlc, sqlx, and GORM do worse than database/sql.

This post is built on the workload, environment, libraries, and methodology in my databases-intuition repo on GitHub. See the repo for details that will help you reproduce or correct me.

continue reading on notes.eatonphil.com

⚠️ This post links to an external website. ⚠️