Skip to content

feat: use parquet for checkpointing #905

Description

@SemyonSinchenko

Is your feature request related to a problem? Please describe.
On my experiments, parquet checkpointing is better for the specific case of "write once, read multiple times". Decoding speed is x1.5-2.0 better and at scale it may be even more.

Image Image Image

Describe the solution you would like

  1. Abstract GraphFramesCheckpointer
  2. Parquet Checkpointing as an alternative (alongside with checkpoint / localCheckpoint)

Component

  • Scala Core Internal
  • Scala API
  • Spark Connect Plugin
  • Infrastructure
  • PySpark Classic
  • PySpark Connect

Additional context

Are you planning on creating a PR?

  • I'm willing to make a pull-request

Activity

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

Metadata

Metadata

Assignees

No one assigned

    Type

    No type

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions