Snowflake bounded source

[turb]

SnowflakeIO has several steps:

1. Run a COPY that outputs partitioned gzipped CSV files in a directory
2. List the directory
3. Read the files
4. Parse CSV

While 1. and 2. are done by one worker, 3. and 4. can be parallelized.

It appears that Google Dataflow is able to do that (using work stealing?), but Apache Flink (with --useDataStreamForBatch=true) propagates the scale of 1. / 2. to 3. 4., leading to very long processing when it can be fully scalable.

This change creates a SnowflakeBoundedSource instead of a simple DoFn to execute the COPY and then read the splits.

When doing that, a bug appears: a race between the apparition of those splits and the reading. It is solved by a change in LazyFlinkSourceSplitEnumerator to make subtasks wait for the splits to be ready.

I tested it still works on Google Dataflow.

---

GitHub Actions Tests Status (on master branch)

See CI.md for more information about GitHub Actions CI or the workflows README to see a list of phrases to trigger workflows.

[gemini-code-assist]

Summary of Changes

Hello, I'm Gemini Code Assist¹! I'm currently reviewing this pull request and will post my feedback shortly. In the meantime, here's a summary to help you and other reviewers quickly get up to speed!

This pull request improves the scalability of SnowflakeIO by introducing a BoundedSource, which allows for more efficient parallel processing of staged CSV files. Additionally, it addresses a race condition in the Flink runner's source split enumeration, ensuring that subtasks correctly wait for splits to become available.

Highlights

• SnowflakeIO Parallelization: Replaced the DoFn-based SnowflakeIO implementation with a BoundedSource to enable better parallelization of file reading after the initial COPY INTO stage.
• Flink Runner Fix: Updated LazyFlinkSourceSplitEnumerator to correctly handle split requests that arrive before the source is fully initialized, preventing race conditions.

🧠 New Feature in Public Preview: You can now enable Memory to help Gemini Code Assist learn from your team's feedback. This makes future code reviews more consistent and personalized to your project's style. Click here to enable Memory in your admin console.

Using Gemini Code Assist

The full guide for Gemini Code Assist can be found on our documentation page, here are some quick tips.

Invoking Gemini

You can request assistance from Gemini at any point by creating a comment using either /gemini <command> or @gemini-code-assist <command>. Below is a summary of the supported commands on the current page.

Feature	Command	Description
Code Review	/gemini review	Performs a code review for the current pull request in its current state.
Pull Request Summary	/gemini summary	Provides a summary of the current pull request in its current state.
Comment	@gemini-code-assist	Responds in comments when explicitly tagged, both in pull request comments and review comments.
Help	/gemini help	Displays a list of available commands.

Customization

To customize Gemini Code Assist for GitHub experience, repository maintainers can create a configuration file and/or provide a custom code review style guide (such as PEP-8 for Python) by creating and adding files to a .gemini/ folder in the base of the repository. Detailed instructions can be found here.

Limitations & Feedback

Gemini Code Assist may make mistakes. Please leave feedback on any instances where its feedback is incorrect or counter productive. You can react with 👍 and 👎 on @gemini-code-assist comments. If you're interested in giving your feedback about your experience with Gemini Code Assist for GitHub and other Google products, sign up here.

Footnotes

1. Review the Privacy Notices, Generative AI Prohibited Use Policy, Terms of Service, and learn how to configure Gemini Code Assist in GitHub here. Gemini can make mistakes, so double check it and use code with caution. ↩

[github-actions]

Assigning reviewers:

R: @ahmedabu98 for label java.

Note: If you would like to opt out of this review, comment assign to next reviewer.

Available commands:

• stop reviewer notifications - opt out of the automated review tooling
• remind me after tests pass - tag the comment author after tests pass
• waiting on author - shift the attention set back to the author (any comment or push by the author will return the attention set to the reviewers)

The PR bot will only process comments in the main thread (not review comments).

[turb]

Hello @Abacn it seems you have implemented a different more thread-safe way of doing the LazyFlinksourceSplitEnumerator part. Thanks!

I suppose I should rebase and get rid of mine, keeping only the Snowflake part?

[Abacn]

Hello @Abacn it seems you have implemented a different more thread-safe way of doing the LazyFlinksourceSplitEnumerator part. Thanks!

I suppose I should rebase and get rid of mine, keeping only the Snowflake part?

Sorry missed that it already been fixed here. Yeah please go ahead

[turb]

@Abacn rebased and tested again to be fixing the original Snowflake issue.

[Abacn]

Dataflow can do it because FileIO.matchAll() has a ReShuffle present by default:

beam/sdks/java/core/src/main/java/org/apache/beam/sdk/io/FileIO.java

Lines 734 to 735 in 6dd599c

	if (getOutputParallelization()) {
	return res.apply(Reshuffle.viaRandomKey());

This introduces a fusion break and the downstream can be parallelized.

It sounds a Flink runner bug that significant impacted its performence. Preferrably a fix should be done in Flink runner and brings back what ReShuffle intended to do (fusion break). Have you tried to not set useDataStreamForBatch (only availlable for Flink 1.x)?

This PR essentially rewrites SnowflakeIO bounded read and would need a closer eye (I could help for generic Java but less experience on Snowflake connecor)

[turb]

We had issues without useDataStreamForBatch. I'll try again on the updated version without it just to be sure. Also we have not tried Flink 2 yet because the support is quite recent.

Before the change, it seems Flink does apply reshuffle, but since it reduced the number of workers to 1, it's reshuffling... in 1 worker, not changing that because the input size (list of files) is tiny.