Skip to content

[Task]: Add perf tests #38559

Open
Open

Description

@chamikaramj

What needs to happen?

For context, please see https://s.apache.org/beam-delta-lake-source.

Issue Priority

Priority: 2 (default / most normal work should be filed as P2)

Issue Components

  • Component: Python SDK
  • Component: Java SDK
  • Component: Go SDK
  • Component: Typescript SDK
  • Component: IO connector
  • Component: Beam YAML
  • Component: Beam examples
  • Component: Beam playground
  • Component: Beam katas
  • Component: Website
  • Component: Infrastructure
  • Component: Spark Runner
  • Component: Flink Runner
  • Component: Prism Runner
  • Component: Twister2 Runner
  • Component: Hazelcast Jet Runner
  • Component: Google Cloud Dataflow Runner

Activity

  1. durgaprasadml commented on May 30, 2026

    @durgaprasadml
    Contributor

    Hi @chamikaramj, I would like to work on this issue.

    I’ve been analyzing the Delta Lake source implementation and the existing Beam IO integration/performance test patterns. My plan is to:

    • Complete the DeltaIO.ReadRows implementation using the Delta Kernel API
    • Add scalable performance/integration tests for Delta Lake reads
    • Reuse existing Beam IOIT utilities and benchmarking patterns
    • Add scenarios for small, large, and partitioned Delta tables
    • Ensure compatibility with Beam runners and existing builder APIs

    I’ll follow the existing Beam IO connector architecture and keep the implementation modular and contribution-ready.

  2. chamikaramj commented on May 30, 2026

    @chamikaramj
    ContributorAuthor

    Sounds good. As discussed in the PR, let's wait for the source implementation with SDF support (#38706) to go in first and then you can add perf testing on top of that.

  3. chamikaramj commented on May 30, 2026

    @chamikaramj
    ContributorAuthor

    For perf testing, let's make sure that that we deploy a real Delta Lake instance with a large input size (for example, 100 GB).

    We should also use a distributed Beam runner instead of a local runner so that the work get parallelized.

    Results can be published to here: https://metrics.beam.apache.org/d/1/getting-started?orgId=1&viewPanel=123125

    Documentation on testing: https://beam.apache.org/documentation/io/testing/

  4. durgaprasadml commented on May 30, 2026

    @durgaprasadml
    Contributor

    I’ll wait for the SDF-based source implementation in #38706 to land first and then continue the perf-testing work on top of that implementation.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions