DataFusion

DataFusion is an extensible query execution framework, written in Rust, that uses Apache Arrow as its in-memory format.

DataFusion supports both an SQL and a DataFrame API for building logical query plans as well as a query optimizer and execution engine capable of parallel execution against partitioned data sources (CSV and Parquet) using threads.

DataFusion also supports distributed query execution via the Ballista crate.

Use Cases

DataFusion is used to create modern, fast and efficient data pipelines, ETL processes, and database systems, which need the performance of Rust and Apache Arrow and want to provide their users the convenience of an SQL interface or a DataFrame API.

Why DataFusion?

High Performance: Leveraging Rust and Arrow's memory model, DataFusion achieves very high performance
Easy to Connect: Being part of the Apache Arrow ecosystem (Arrow, Parquet and Flight), DataFusion works well with the rest of the big data ecosystem
Easy to Embed: Allowing extension at almost any point in its design, DataFusion can be tailored for your specific usecase
High Quality: Extensively tested, both by itself and with the rest of the Arrow ecosystem, DataFusion can be used as the foundation for production systems.

Known Uses

Here are some of the projects known to use DataFusion:

Ballista Distributed Compute Platform
Cloudfuse Buzz
Cube Store
datafusion-python
delta-rs
InfluxDB IOx Time Series Database
ROAPI
Tensorbase
Squirtle

(if you know of another project, please submit a PR to add a link!)

Example Usage

Run a SQL query against data stored in a CSV:

use datafusion::prelude::*;
use arrow::util::pretty::print_batches;
use arrow::record_batch::RecordBatch;

#[tokio::main]
async fn main() -> datafusion::error::Result<()> {
  // register the table
  let mut ctx = ExecutionContext::new();
  ctx.register_csv("example", "tests/example.csv", CsvReadOptions::new())?;

  // create a plan to run a SQL query
  let df = ctx.sql("SELECT a, MIN(b) FROM example GROUP BY a LIMIT 100")?;

  // execute and print results
  let results: Vec<RecordBatch> = df.collect().await?;
  print_batches(&results)?;
  Ok(())
}

Use the DataFrame API to process data stored in a CSV:

use datafusion::prelude::*;
use arrow::util::pretty::print_batches;
use arrow::record_batch::RecordBatch;

#[tokio::main]
async fn main() -> datafusion::error::Result<()> {
  // create the dataframe
  let mut ctx = ExecutionContext::new();
  let df = ctx.read_csv("tests/example.csv", CsvReadOptions::new())?;

  let df = df.filter(col("a").lt_eq(col("b")))?
          .aggregate(vec![col("a")], vec![min(col("b"))])?
          .limit(100)?;

  // execute and print results
  let results: Vec<RecordBatch> = df.collect().await?;
  print_batches(&results)?;
  Ok(())
}

Both of these examples will produce

+---+--------+
| a | MIN(b) |
+---+--------+
| 1 | 2      |
+---+--------+

Using DataFusion as a library

DataFusion is published on crates.io, and is well documented on docs.rs.

To get started, add the following to your Cargo.toml file:

[dependencies]
datafusion = "4.0.0-SNAPSHOT"

Using DataFusion as a binary

DataFusion also includes a simple command-line interactive SQL utility. See the CLI reference for more information.

Status

General

SQL Support

Data Sources

CSV
Parquet primitive types
Parquet nested types

Extensibility

DataFusion is designed to be extensible at all points. To that end, you can provide your own custom:

User Defined Functions (UDFs)
User Defined Aggregate Functions (UDAFs)
User Defined Table Source (TableProvider) for tables
User Defined Optimizer passes (plan rewrites)
User Defined LogicalPlan nodes
User Defined ExecutionPlan nodes

Supported SQL

This library currently supports many SQL constructs, including

CREATE EXTERNAL TABLE X STORED AS PARQUET LOCATION '...'; to register a table's locations
SELECT ... FROM ... together with any expression
ALIAS to name an expression
CAST to change types, including e.g. Timestamp(Nanosecond, None)
most mathematical unary and binary expressions such as +, /, sqrt, tan, >=.
WHERE to filter
GROUP BY together with one of the following aggregations: MIN, MAX, COUNT, SUM, AVG
ORDER BY together with an expression and optional ASC or DESC and also optional NULLS FIRST or NULLS LAST

Supported Functions

DataFusion strives to implement a subset of the PostgreSQL SQL dialect where possible. We explicitly choose a single dialect to maximize interoperability with other tools and allow reuse of the PostgreSQL documents and tutorials as much as possible.

Currently, only a subset of the PostgreSQL dialect is implemented, and we will document any deviations.

Schema Metadata / Information Schema Support

DataFusion supports the showing metadata about the tables available. This information can be accessed using the views of the ISO SQL information_schema schema or the DataFusion specific SHOW TABLES and SHOW COLUMNS commands.

More information can be found in the Postgres docs).

To show tables available for use in DataFusion, use the SHOW TABLES command or the information_schema.tables view:

> show tables;
+---------------+--------------------+------------+------------+
| table_catalog | table_schema       | table_name | table_type |
+---------------+--------------------+------------+------------+
| datafusion    | public             | t          | BASE TABLE |
| datafusion    | information_schema | tables     | VIEW       |
+---------------+--------------------+------------+------------+

> select * from information_schema.tables;

+---------------+--------------------+------------+--------------+
| table_catalog | table_schema       | table_name | table_type   |
+---------------+--------------------+------------+--------------+
| datafusion    | public             | t          | BASE TABLE   |
| datafusion    | information_schema | TABLES     | SYSTEM TABLE |
+---------------+--------------------+------------+--------------+

To show the schema of a table in DataFusion, use the SHOW COLUMNS command or the or information_schema.columns view:

> show columns from t;
+---------------+--------------+------------+-------------+-----------+-------------+
| table_catalog | table_schema | table_name | column_name | data_type | is_nullable |
+---------------+--------------+------------+-------------+-----------+-------------+
| datafusion    | public       | t          | a           | Int32     | NO          |
| datafusion    | public       | t          | b           | Utf8      | NO          |
| datafusion    | public       | t          | c           | Float32   | NO          |
+---------------+--------------+------------+-------------+-----------+-------------+

>   select table_name, column_name, ordinal_position, is_nullable, data_type from information_schema.columns;
+------------+-------------+------------------+-------------+-----------+
| table_name | column_name | ordinal_position | is_nullable | data_type |
+------------+-------------+------------------+-------------+-----------+
| t          | a           | 0                | NO          | Int32     |
| t          | b           | 1                | NO          | Utf8      |
| t          | c           | 2                | NO          | Float32   |
+------------+-------------+------------------+-------------+-----------+

Supported Data Types

DataFusion uses Arrow, and thus the Arrow type system, for query execution. The SQL types from sqlparser-rs are mapped to Arrow types according to the following table

SQL Data Type	Arrow DataType
`CHAR`	`Utf8`
`VARCHAR`	`Utf8`
`UUID`	Not yet supported
`CLOB`	Not yet supported
`BINARY`	Not yet supported
`VARBINARY`	Not yet supported
`DECIMAL`	`Float64`
`FLOAT`	`Float32`
`SMALLINT`	`Int16`
`INT`	`Int32`
`BIGINT`	`Int64`
`REAL`	`Float64`
`DOUBLE`	`Float64`
`BOOLEAN`	`Boolean`
`DATE`	`Date32`
`TIME`	`Time64(TimeUnit::Millisecond)`
`TIMESTAMP`	`Date64`
`INTERVAL`	Not yet supported
`REGCLASS`	Not yet supported
`TEXT`	Not yet supported
`BYTEA`	Not yet supported
`CUSTOM`	Not yet supported
`ARRAY`	Not yet supported

Architecture Overview

There is no formal document describing DataFusion's architecture yet, but the following presentations offer a good overview of its different components and how they interact together.

(March 2021): The DataFusion architecture is described in Query Engine Design and the Rust-Based DataFusion in Apache Arrow: recording (DataFusion content starts ~ 15 minutes in) and slides
(Feburary 2021): How DataFusion is used within the Ballista Project is described in *Ballista: Distributed Compute with Rust and Apache Arrow: recording

Developer's guide

Please see Developers Guide for information about developing DataFusion.

kszucs / arrow-datafusion