Member-only story

I Tried 5 Repository Patterns in Go — Only One Survived

I tested every Go repository pattern on real projects — here is the brutal honest result of what actually works.

10 min readJul 23, 2026

I have built a lot of Go services. Small ones, large ones, ones that started clean and slowly turned into something I am not proud of. And across all of them, the question that caused the most debate, the most rewrites, and the most late-night Slack arguments was always the same one.

How do we structure our data access layer?

It sounds boring. It is not. Get it wrong and you end up with a codebase where adding a new filter to a query requires touching six files. Get it right and your data layer becomes the most boring, reliable part of your entire system — which is exactly what you want.

Over the years I have tried five different approaches to the repository pattern in Go. Each one taught me something. Only one of them has survived every project I have brought it to without eventually causing pain.

Let me walk you through all five — the good, the bad, and the one that actually works.

— -

Pattern 1: The Struct With Methods (The Beginner’s Choice)

This is a masterful breakdown of the repository pattern in Go—separating domain-specific lookups from flexible search filters is such a clean, pragmatic balance. The rule about naming queries after domain concepts vs. using filters is definitely a keeper for future architectural decisions!

I think I have been through every single one of what you mention, but I end up with "pattern 6" which is just CQS in Go, I dont mock anything, I just have a different databases for tests, run every package tests inside a transaction and run tests…

I Tried 5 Repository Patterns in Go — Only One Survived

I tested every Go repository pattern on real projects — here is the brutal honest result of what actually works.

Pattern 1: The Struct With Methods (The Beginner’s Choice)

Written by Ade Mawan

Responses (3)

More from Ade Mawan

I Stopped Writing db: Tags in Go — Here’s What I Use Instead

sqlc generates type-safe Go code from your SQL queries — no struct tags, no manual scanning, no runtime mismatches.

The Cost of Abstraction: When to Break SOLID Rules in High-Performance Go

We obsess over clean architecture — and pay the price in latency. Here is when to choose raw speed instead.

I Tried 3 ORMs in Go (and No ORM) — Here’s the Only One I’d Recommend

I benchmarked GORM, ent, sqlx, and raw SQL in Go — here’s the ORM that actually earned my trust, and why.

Interfaces Everywhere vs Interfaces Where Needed: A Go Architecture Experiment

A hands-on Go experiment comparing interfaces-everywhere against interfaces-where-needed — the results might change how you code.

Recommended from Medium

AI Isn’t Replacing Developers. It’s Doing Something Worse.

The replacement narrative is loud. The quiet one is already happening.

Andrej Karpathy’s New suggestion- Ask AI output in ASD-STE100 format and not in Text format

A few days ago, Andrej Karpathy shared a perspective that immediately clicked with how my own daily workflow has been evolving. His premise…

So Good News，Go 1.28 Finally Introduces Inferred Composite Literals: A Proposal That Took 11 Years

If you’re a gopher who has been writing Go for a long time, you’ve definitely had this frustrating moment: the compiler knows exactly what…

High-Performance JSON Streaming in Golang

When developers compare JSON parsers, the discussion usually becomes predictable very quickly.

Go Env Configuration Done Right

Part of the Go Production Series — a collection of opinionated, battle-tested patterns for building Go services in the real world.

I Stopped Writing db: Tags in Go — Here’s What I Use Instead

sqlc generates type-safe Go code from your SQL queries — no struct tags, no manual scanning, no runtime mismatches.

Responses (3)

More from Ade Mawan

I Stopped Writing db: Tags in Go — Here’s What I Use Instead

sqlc generates type-safe Go code from your SQL queries — no struct tags, no manual scanning, no runtime mismatches.

The Cost of Abstraction: When to Break SOLID Rules in High-Performance Go

We obsess over clean architecture — and pay the price in latency. Here is when to choose raw speed instead.

I Tried 3 ORMs in Go (and No ORM) — Here’s the Only One I’d Recommend

I benchmarked GORM, ent, sqlx, and raw SQL in Go — here’s the ORM that actually earned my trust, and why.

Interfaces Everywhere vs Interfaces Where Needed: A Go Architecture Experiment

A hands-on Go experiment comparing interfaces-everywhere against interfaces-where-needed — the results might change how you code.

Recommended from Medium

AI Isn’t Replacing Developers. It’s Doing Something Worse.

The replacement narrative is loud. The quiet one is already happening.

Andrej Karpathy’s New suggestion- Ask AI output in ASD-STE100 format and not in Text format

A few days ago, Andrej Karpathy shared a perspective that immediately clicked with how my own daily workflow has been evolving. His premise…

So Good News，Go 1.28 Finally Introduces Inferred Composite Literals: A Proposal That Took 11 Years

If you’re a gopher who has been writing Go for a long time, you’ve definitely had this frustrating moment: the compiler knows exactly what…

High-Performance JSON Streaming in Golang

When developers compare JSON parsers, the discussion usually becomes predictable very quickly.

Go Env Configuration Done Right

Part of the Go Production Series — a collection of opinionated, battle-tested patterns for building Go services in the real world.

I Stopped Writing db: Tags in Go — Here’s What I Use Instead

sqlc generates type-safe Go code from your SQL queries — no struct tags, no manual scanning, no runtime mismatches.