hudi-rs icon indicating copy to clipboard operation
hudi-rs copied to clipboard

Adopt OpenDAL for accessing storage layer

Open alberttwong opened this issue 1 year ago • 6 comments

Is there an existing issue for this?

  • [X] I have searched the existing issues

Description of the bug

OpenDAL offers a unified data access layer, empowering users to seamlessly and efficiently retrieve data from diverse storage services. Our goal is to deliver a comprehensive solution for any languages, methods, integrations, and services.

ParadeDB uses Apache OpenDAL to integrate with object stores. If we support OpenDAL, all databases that use OpenDAL will be able to use Hudi

Steps To Reproduce

n/a

Expected behavior

n/a

Screenshots / Logs

No response

Software information

n/a

Additional context

No response

alberttwong avatar Sep 03 '24 19:09 alberttwong

https://github.com/apache/opendal/issues/5065

alberttwong avatar Sep 03 '24 19:09 alberttwong

Thank you for bringing this up. I'm happy to help.

Xuanwo avatar Sep 09 '24 13:09 Xuanwo

there's a crate that brings opendal integrations to object_store hudi-rs uses: https://crates.io/crates/object_store_opendal It may be worth considering this instead of replacing it with opendal completely?

kazdy avatar Sep 22 '24 09:09 kazdy

@Xuanwo and I had a chat - we don't have to bridge the APIs as the current object_store footprint is still small within the codebase - the API migration won't be too much of changes. And we should make use of opendal's capabilities to support more storage systems natively.

xushiyan avatar Nov 27 '24 18:11 xushiyan

Hi, @xushiyan and @alberttwong, I'm working on this now. One blocker here is opendal doesn't support construct from url yet, I will add this support in opendal first.

Tracked at https://github.com/apache/opendal/issues/5407

Xuanwo avatar Dec 11 '24 09:12 Xuanwo

Waiting for https://github.com/apache/opendal/pull/5482 and a new release that includes it.

xushiyan avatar Mar 21 '25 16:03 xushiyan