Databend vs. Snowflake Features and Capabilities

[BohuTANG]

Summary

Databend is an open-source alternative to Snowflake, so we need a page to introduce the differences between Databend and Snowflake.
Feel free to contribute if you notice any omissions.

	Category	Snowflake	Databend
Numeric Data Types	Data Types	Yes	Yes [1, 2]
String & Binary Data Types	Data Types	Yes	Yes
Logical Data Types	Data Types	Yes	Yes
Date & Time Data Types	Data Types	Yes	Yes
Geospatial Data Types	Data Types	Yes	WIP
Bitmap Data Types	Data Types	No	Yes
Semi-structured Data Types	Data Types	Yes	Yes
Database	Databases	Yes	Yes
Table	Tables	Yes	Yes
External Table	Tables	Yes	Yes
Vacuum	Tables	No	Yes
Clustering	Tables	Yes	Yes(Distributed)
Full-Text Index	Tables	No	Yes
Hive	Lake	Yes	Yes
Iceberg	Lake	Yes(Private Preview)	Yes
View	Views	Yes	Yes
Materialized Views	Views	Yes	Yes
User, Role, & Privilege	Privileges	Yes	Yes
Query	Queries	Yes	Yes
Update	General DML	Yes	Yes(Distributed)
Replace	General DML	No	Yes(Distributed)
Delete	General DML	Yes	Yes(Distributed)
Merge	General DML	Yes	Yes
Stage(Internal/External/Remove/List)	Data Loading	Yes	Yes
Query Stage File	Data Loading	Yes	Yes
Copy Into Table/Location	Data Loading	Yes	Yes(Distributed)
Transforming Data During a Load	Data Loading	Yes	Yes
Table Pipe	Data Loading	Yes	Yes(Cloud)
Table Task	Data Loading	Yes	Yes
Table Stream(Append-only/Standard)	Data Loading	Yes	Yes
User-Defined Functions	Functions	Yes	Yes
External Functions	Functions	Yes	Yes
Stored Procedure	Functions	Yes	Planning
AI Functions	Functions	No	Yes
Window Function	Functions	Yes	Yes
Time Travel	Data Recovery	Yes	Yes
Data Sharing	Data Sharing	Yes	Yes
RBAC	Security	Yes	Yes
Network Policies	Security	Yes	Yes
Masking Policies	Security	Yes	Yes
Kafka Connect	Connectors	Yes	Yes
Flink CDC	Connectors	No	Yes
dbt	Connectors	Yes	Yes
Drivers	Drivers	Go Java Ptyhon Node.js	Go Java Python Node.js Rust
Embedded Python Library		No	Yes
Storage Format	Storage	MicroPartition	Parquet
Aggregate&Join&Sort Spill	Performance Optimization	Yes	Yes
Local Disk Cache	Performance Optimization	Yes	Yes
Profiling	Performance Optimization	Powerful	Powerful
CLI Client	Client	SnowSQL	BendSQL
TPC-H	Benchmark	Yes	Yes
TPC-DS	Benchmark	Yes	Yes
Supported Cloud Platforms	Cloud	AWS/GCP/AZURE	Any Cloud

[ZhiHanZ]

Feature	Snowflake	Databend
Kafka Connect	Yes	Yes
Flink CDC	No	Yes
dbt	Yes	Yes
go driver	Yes	Yes
java driver	Yes	Yes
rust driver	No	Yes
python driver	Yes	Yes
nodejs driver	Yes	Planning
embedded python lib	No	Yes
Iceberg	Private Preview	Yes

[PsiACE]

we hava a nodejs driver, see:

• https://www.npmjs.com/package/databend-driver
• https://databend.rs/doc/develop/nodejs

In addition, although we do not yet support geospatial types, we have already implemented a wide range of geospatial functions. I believe this is worth mentioning.

• https://databend.rs/doc/sql-functions/geo-functions/

I think, in Storage Format, perhaps we can also add our native format and indicate that it can bring better performance.

[xudong963]

We also support aggregate and hash join spilling.

[Chasen-Zhang]

Summary

Databend is an open-source alternative to Snowflake, so we need a page to introduce the differences between Databend and Snowflake. Feel free to contribute if you notice any omissions.

Category Snowflake Databend
Numeric Data Types Data Types Yes Yes [1, 2]
String & Binary Data Types Data Types Yes Yes
Logical Data Types Data Types Yes Yes
Date & Time Data Types Data Types Yes Yes
Geospatial Data Types Data Types Yes No(Geo functions)
Bitmap Data Types Data Types No Yes
Semi-structured Data Types Data Types Yes Yes
Database Databases Yes Yes
Table Tables Yes Yes
External Table Tables Yes Yes
Vacuum Tables No Yes
Clustering Tables Yes Yes(Distributed)
Hive Lake Yes Yes
Iceberg Lake Yes(Private Preview) Yes
View Views Yes Yes
Materialized Views Views Yes Yes(Need Doc)
User, Role, & Privilege Privileges Yes Yes
Query Queries Yes Yes
Update General DML Yes Yes(Distributed)
Replace General DML No Yes(Distributed)
Delete General DML Yes Yes(Distributed)
Merge General DML Yes Yes
Stage(Internal/External/Remove/List) Data Loading Yes Yes
Query Stage File Data Loading Yes Yes
Copy Into Table/Location Data Loading Yes Yes(Distributed)
Transforming Data During a Load Data Loading Yes Yes
Table Pipe Data Loading Yes Yes(Cloud)
Table Task Data Loading Yes Yes
Table Stream Data Loading Yes Yes
User-Defined Functions Functions Yes Yes
External Functions Functions Yes Yes
Stored Procedure Functions Yes No
AI Functions Functions No Yes
Window Function Functions Yes Yes
Time Travel Data Recovery Yes Yes
Data Sharing Data Sharing Yes Yes
RBAC Security Yes Yes
Network Policies Security Yes Yes
Masking Policies Security Yes Yes
Kafka Connect Connectors Yes Yes
Flink CDC Connectors No Yes
dbt Connectors Yes Yes
Drivers Drivers Go
Java
Ptyhon
Node.js Go
Java
Python
Node.js
Rust
Embedded Python Library No Yes
Storage Format Storage MicroPartition Parquet
Aggregate&Join Spill Performance Optimization Yes Yes
Local Disk Cache Performance Optimization Yes Yes
Profiling Performance Optimization Powerful Medium
CLI Client Client SnowSQL BendSQL
TPC-H Benchmark Yes Yes
TPC-DS Benchmark Yes Yes
Supported Cloud Platforms Cloud AWS/GCP/AZURE Any Cloud

@BohuTANG Could you help me change the address of the connection，
Here are the rules：
/doc/sql-reference/* => /sql/sql-reference/*
/doc/sql-commands/* => /sql/sql-commands/*
/doc/sql-functions/* =>/sql/sql-functions/*
(👆 Although I redirected all of these, it's better to change them all together. And you can directly change the domain name to https://docs.databend.com )

https://docs.databend.com/load-data/pipeline => https://docs.databend.com/doc/load-data/pipeline
https://databend.rs/blog/2023-08-01-iceberg-integration=> https://www.databend.com/blog/2023-08-01-iceberg-integration/

Thank you in advance！

[cdmikechen]

@BohuTANG
Wish to support Geospatial Data Types, we used to have a lot of spatial data needs, and if databend can support geo, it will be a great help to us ~