Google Git
Sign in
apache / spark / cc3df38ce4b76d3a7d5578ac5824afe498ed59fa
commitcc3df38ce4b76d3a7d5578ac5824afe498ed59fa[log] [tgz]
authorLuca Canali <luca.canali@cern.ch>Wed Sep 04 13:52:49 2024 -0700
committerallisonwang-db <allison.wang@databricks.com>Wed Sep 04 13:52:49 2024 -0700
tree3c4f75bf18d210df46f1a2bb8e4f668125945380
parent2ed6c3e511f322c5fd01953736c376a85ff2c687 [diff]
[SPARK-48493][PYTHON] Enhance Python Datasource Reader with direct Arrow Batch support for improved performance

### What changes were proposed in this pull request?
This pull request proposes enhancing the Python Datasource Reader by adding an option to yield Arrow batches directly. This change aims to significantly improve performance compared to the existing approach of using tuples or Rows. The implementation takes advantage of the existing work with MapInArrow (referenced in SPARK-46253).

### Why are the changes needed?
The changes are needed to address performance issues in the Python Datasource Reader. The current method of sending data as tuples or Rows is inefficient, leading to slower data processing times. By allowing the Datasource Reader to yield Arrow batches directly, we can use the more efficient Arrow format, significantly speeding up data processing. Tests have shown this approach to be up to 8x faster (in a preliminary test with a High Energy Physics Datasource reader for the ROOT data format), particularly benefiting use cases involving large datasets.

### Does this PR introduce _any_ user-facing change?
Yes, this PR introduces a user-facing change by adding an option to the Python Datasource Reader that allows users to yield Arrow batches directly.

### How was this patch tested?
A new test was added to the Python Datasource test suite. Additionally, it was manually tested using a custom Python datasource for performance testing.

### Was this patch authored or co-authored using generative AI tooling?
No

Closes #46826 from LucaCanali/arrowBatchesInPythonDatasource.

Authored-by: Luca Canali <luca.canali@cern.ch>
Signed-off-by: allisonwang-db <allison.wang@databricks.com>
5 files changed
tree: 3c4f75bf18d210df46f1a2bb8e4f668125945380
  1. .github/
  2. assembly/
  3. bin/
  4. binder/
  5. build/
  6. common/
  7. conf/
  8. connector/
  9. core/
  10. data/
  11. dev/
  12. docs/
  13. examples/
  14. graphx/
  15. hadoop-cloud/
  16. launcher/
  17. licenses/
  18. licenses-binary/
  19. mllib/
  20. mllib-local/
  21. project/
  22. python/
  23. R/
  24. repl/
  25. resource-managers/
  26. sbin/
  27. sql/
  28. streaming/
  29. tools/
  30. ui-test/
  31. .asf.yaml
  32. .gitattributes
  33. .gitignore
  34. CONTRIBUTING.md
  35. LICENSE
  36. LICENSE-binary
  37. NOTICE
  38. NOTICE-binary
  39. pom.xml
  40. README.md
  41. scalastyle-config.xml
README.md

Apache Spark

Spark is a unified analytics engine for large-scale data processing. It provides high-level APIs in Scala, Java, Python, and R, and an optimized engine that supports general computation graphs for data analysis. It also supports a rich set of higher-level tools including Spark SQL for SQL and DataFrames, pandas API on Spark for pandas workloads, MLlib for machine learning, GraphX for graph processing, and Structured Streaming for stream processing.

https://spark.apache.org/

GitHub Actions Build PySpark Coverage PyPI Downloads

Online Documentation

You can find the latest Spark documentation, including a programming guide, on the project web page. This README file only contains basic setup instructions.

Building Spark

Spark is built using Apache Maven. To build Spark and its example programs, run:

./build/mvn -DskipTests clean package

(You do not need to do this if you downloaded a pre-built package.)

More detailed documentation is available from the project site, at “Building Spark”.

For general development tips, including info on developing Spark using an IDE, see “Useful Developer Tools”.

Interactive Scala Shell

The easiest way to start using Spark is through the Scala shell:

./bin/spark-shell

Try the following command, which should return 1,000,000,000:

scala> spark.range(1000 * 1000 * 1000).count()

Interactive Python Shell

Alternatively, if you prefer Python, you can use the Python shell:

./bin/pyspark

And run the following command, which should also return 1,000,000,000:

>>> spark.range(1000 * 1000 * 1000).count()

Example Programs

Spark also comes with several sample programs in the examples directory. To run one of them, use ./bin/run-example <class> [params]. For example:

./bin/run-example SparkPi

will run the Pi example locally.

You can set the MASTER environment variable when running examples to submit examples to a cluster. This can be spark:// URL, “yarn” to run on YARN, and “local” to run locally with one thread, or “local[N]” to run locally with N threads. You can also use an abbreviated class name if the class is in the examples package. For instance:

MASTER=spark://host:7077 ./bin/run-example SparkPi

Many of the example programs print usage help if no params are given.

Running Tests

Testing first requires building Spark. Once Spark is built, tests can be run using:

./dev/run-tests

Please see the guidance on how to run tests for a module, or individual tests.

There is also a Kubernetes integration test, see resource-managers/kubernetes/integration-tests/README.md

A Note About Hadoop Versions

Spark uses the Hadoop core library to talk to HDFS and other Hadoop-supported storage systems. Because the protocols have changed in different versions of Hadoop, you must build Spark against the same version that your cluster runs.

Please refer to the build documentation at “Specifying the Hadoop Version and Enabling YARN” for detailed guidance on building for a particular distribution of Hadoop, including building for particular Hive and Hive Thriftserver distributions.

Configuration

Please refer to the Configuration Guide in the online documentation for an overview on how to configure Spark.

Contributing

Please review the Contribution to Spark guide for information on how to get started contributing to the project.