GitHub user jianlirong added a comment to the discussion: [Proposal] Remove 
`gpcontrib/gpmapreduce` starting from Cloudberry 3.0

I agree with this proposal.

>From our perspective, `gpmapreduce` is a legacy component with very limited 
>usage today, and we are not aware of any significant active user base relying 
>on it.

More importantly, the data ecosystem has changed considerably since 
`gpmapreduce` was originally introduced. With the growing adoption of the 
Lakehouse architecture, there are now many more mature approaches for data 
interoperability between MPP databases and big data platforms, including 
external tables, FDWs, open table formats, and various compute/storage 
integration solutions.

Compared with these newer approaches, `gpmapreduce` does not have any obvious 
advantage, either in terms of performance or user adoption. Its original use 
cases can generally be addressed more effectively through modern data 
integration and interoperability mechanisms.

I also think this is an important part of keeping Apache Cloudberry moving 
forward. As an actively evolving open-source project, we need to not only 
continuously add support for new technologies and emerging use cases, but also 
periodically remove legacy components that no longer provide sufficient value. 
Otherwise, the codebase will inevitably become more complex and the maintenance 
burden will continue to grow.

This is a common pattern in mature open-source infrastructure projects. For 
example, Linux continues to add support for new hardware architectures and 
platforms while also removing support for obsolete hardware and platforms when 
they are no longer widely used. This allows the project to evolve without 
carrying unnecessary legacy code indefinitely.

In that sense, removing `gpmapreduce` is not just about removing an old 
feature; it is part of keeping Cloudberry's architecture and codebase focused 
on technologies that are relevant to its current and future users.

Therefore, I support removing `gpcontrib/gpmapreduce` starting from Cloudberry 
3.0. Keeping it in the 2.x line for existing users, as proposed, also seems 
like a reasonable migration path.

Lirong

GitHub link: 
https://github.com/apache/cloudberry/discussions/2030#discussioncomment-18522374

----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to