goorm

Continuous Benchmark for ORM libraries written in golang.

English | 中文

Overview

This repository provides performance benchmarks for popular Go ORM libraries using SQLite as the test database. The benchmarks are designed to compare the performance of different ORM libraries across common database operations.

Tested ORM Libraries

• ZORM - A simple, ultra-fast ORM library
• BORM - A better ORM library that is simple, fast and self-mockable for Go
• GORM - The most popular ORM library for Go
• SQLX - A library which provides a set of extensions on go's standard database/sql library
• XORM - A simple and powerful ORM for Go
• ENT - An entity framework for Go
• BUN - SQL-first Golang ORM

Benchmark Tests

The following operations are benchmarked:

Test Case	Description
InsertSingle	Single record insertion performance
InsertBatch	Batch insertion performance (100 records per batch)
GetByID	Single record retrieval by primary key
GetByIDs	Multiple records retrieval by primary keys
Update	Record update performance
Delete	Record deletion performance
Count	Count query performance
GetAll	Paginated query performance (limit/offset)

Running Benchmarks

Prerequisites

• Go 1.21 or higher
• SQLite3

Run All Benchmarks

go test -bench=. -benchmem

Run Specific ORM Benchmark

# ZORM only
go test -bench=ZORM -benchmem

# BORM only
go test -bench=BORM -benchmem

# GORM only
go test -bench=GORM -benchmem

# XORM only
go test -bench=XORM -benchmem

# SQLX only
go test -bench=SQLX -benchmem

# BUN only
go test -bench=BUN -benchmem

# ENT only
go test -bench=ENT -benchmem

Run Specific Test Case

# Insert single test for all ORMs
go test -bench=InsertSingle -benchmem

# GetByID test for all ORMs
go test -bench=GetByID -benchmem

Benchmark Results

Quick Summary

Test Case	ZORM	BORM	BUN	ENT	GORM	SQLX	XORM
InsertSingle	🟢 1x	🟢 1x	🟠 3.13x	🟠 3.46x	🔴 7.09x	🔴 60.61x	🔴 61.12x
InsertBatch	🟢 1x	🟢 1x	🟡 1.30x	🟠 2.50x	🟡 1.89x	🟠 3.57x	🟠 3.33x
GetByID	🟢 1x	🟢 1x	🟡 1.52x	🟡 1.85x	🟡 1.90x	🟠 2x	🟠 3.12x
GetByIDs	🟢 1x	🟢 1x	🟡 1.17x	🟡 1.38x	🟡 1.39x	🟡 1.36x	🟡 1.98x
Update	🟢 1x	🟢 1x	🟠 2.67x	🔴 9.86x	🔴 7.06x	🔴 82.52x	🔴 84x
Delete	🟢 1x	🟢 1x	🟠 2.31x	🟠 2.62x	🔴 6.40x	🔴 105.84x	🔴 101.85x
Count	🟢 1x	🟢 1x	🟠 2.15x	🔴 13.40x	🟠 2.99x	🟠 4.34x	🔴 5.95x
GetAll	🟢 1x	🟢 1x	🟡 1.14x	🟡 1.21x	🟡 1.43x	🟡 1.18x	🟡 1.91x

Ratio indicates performance multiplier relative to the fastest ORM (lower is better)

⭐ indicates the ORM is both fast and memory-efficient for this test case (Pareto-optimal in ns/op and B/op, lower is better). Stars are placed in the ns/op and B/op columns.

Detailed Results

Test Environment

• Go Version: 1.21+
• Database: SQLite (in-memory, DSN uses cache=shared&mode=memory)
• CPU: M4 Pro
• OS: darwin (amd64)

The benchmark results show:

• ns/op: Nanoseconds per operation
• B/op: Bytes allocated per operation
• allocs/op: Number of allocations per operation

InsertSingle (Single Record Insertion)

ORM	ns/op	Ratio	B/op	allocs/op
ZORM	2,913 ⭐	🟢 1x	464 ⭐	16
BORM	2,916	🟢 1x	464	16
BUN	9,123	🟠 3.13x	5,405	27
ENT	10,065	🟠 3.46x	2,642	74
GORM	20,652	🔴 7.09x	6,116	96
SQLX	176,553	🔴 60.61x	632	18
XORM	178,033	🔴 61.12x	2,690	54

InsertBatch (Batch Insertion - 100 records)

ORM	ns/op	Ratio	B/op	allocs/op
BORM	104,405	🟢 1x	59,503	912
ZORM	104,862	🟢 1x	59,502 ⭐	912
BUN	136,110	🟡 1.30x	24,455	723
GORM	197,739	🟡 1.89x	74,929	1,494
ENT	261,124	🟠 2.50x	213,805	3,360
XORM	347,846	🟠 3.33x	98,830	2,449
SQLX	373,004	🟠 3.57x	47,349	1,622

GetByID (Single Record Retrieval)

ORM	ns/op	Ratio	B/op	allocs/op
ZORM	4,729 ⭐	🟢 1x	939 ⭐	33
BORM	4,776	🟢 1x	939	33
BUN	7,191	🟡 1.52x	5,700	36
ENT	8,766	🟡 1.85x	3,812	103
GORM	8,975	🟡 1.90x	4,076	73
SQLX	9,471	🟠 2x	1,155	37
XORM	14,750	🟠 3.12x	4,809	139

GetByIDs (Multiple Records Retrieval - 10 records)

ORM	ns/op	Ratio	B/op	allocs/op
BORM	18,082 ⭐	🟢 1x	3,511 ⭐	95
ZORM	18,257	🟢 1x	3,511	95
BUN	21,171	🟢 1.17x	7,467	107
SQLX	24,623	🟡 1.36x	4,051	116
ENT	24,997	🟡 1.38x	9,679	230
GORM	25,137	🟡 1.39x	7,369	186
XORM	35,883	🟡 1.98x	12,918	400

Update (Record Update)

ORM	ns/op	Ratio	B/op	allocs/op
ZORM	2,103 ⭐	🟢 1x	454 ⭐	13
BORM	2,107	🟢 1x	454	13
BUN	5,607	🟠 2.67x	5,044	15
GORM	14,845	🔴 7.06x	7,442	99
ENT	20,745	🔴 9.86x	5,608	156
SQLX	173,533	🔴 82.52x	654	16
XORM	176,658	🔴 84x	4,082	103

Delete (Record Deletion)

ORM	ns/op	Ratio	B/op	allocs/op
BORM	1,646	🟢 1x	160	7
ZORM	1,667	🟢 1x	159 ⭐	7
BUN	3,798	🟠 2.31x	4,880	14
ENT	4,318	🟠 2.62x	1,832	44
GORM	10,536	🔴 6.40x	5,571	75
XORM	167,642	🔴 101.85x	3,043	80
SQLX	174,217	🔴 105.84x	255	10

Count (Count Query)

ORM	ns/op	Ratio	B/op	allocs/op
ZORM	1,546 ⭐	🟢 1x	440 ⭐	14
BORM	1,555	🟢 1x	440	14
BUN	3,327	🟠 2.15x	1,288	23
GORM	4,617	🟠 2.99x	2,720	33
SQLX	6,716	🟠 4.34x	504	16
XORM	9,193	🔴 5.95x	2,394	61
ENT	20,720	🔴 13.40x	2,384	54

GetAll (Paginated Query - limit 100, offset)

ORM	ns/op	Ratio	B/op	allocs/op
ZORM	95,365 ⭐	🟢 1x	14,829 ⭐	607
BORM	95,795	🟢 1x	14,829	607
BUN	109,081	🟢 1.14x	21,955	709
SQLX	112,556	🟢 1.18x	17,526	712
ENT	115,475	🟡 1.21x	43,301	1,256
GORM	136,701	🟡 1.43x	25,311	1,128
XORM	182,305	🟡 1.91x	74,869	2,771

Test Database

All benchmarks use SQLite as the test database:

• In-memory database for fast performance
• Temporary files are automatically cleaned up after tests
• Each ORM uses its own isolated database instance

Project Structure

goorm/
├── gorm/          # GORM implementation
├── xorm/          # XORM implementation
├── zorm/          # ZORM implementation
├── sqlx/           # SQLX implementation
├── borm/           # BORM implementation
├── bun/            # BUN implementation
├── ent/             # ENT implementation
│   └── schema/      # ENT schema definitions
├── internal/
│   ├── models/     # Test models (User, Post)
│   └── orm/        # Unified ORM interface
├── goorm_test.go   # Benchmark tests
├── go.mod          # Go module file
└── README.md       # This file

Adding New ORM Libraries

To add a new ORM library:

1. Create a new directory (e.g., ent/)
2. Implement the ORMInterface in a new file
3. Add the ORM to the orms map in goorm_test.go
4. Add benchmark functions following the naming pattern

Note for ENT: ENT requires code generation. After adding the schema files, run:

go generate ./ent

This will generate the ENT client code needed for the implementation.

Contributing

Contributions are welcome! Please feel free to submit a Pull Request.

License

MIT License - see LICENSE file for details.

References

• Inspired by benchplus/gocache
• ZORM Repository
• BORM Repository
• GORM Documentation
• XORM Documentation
• SQLX Documentation
• ENT Documentation
• BUN Documentation