Blog: Using Ordering for Better Plans in Apache DataFusion

[akurmustafa]

As per the discussion. I am moving some of my previous blog posts to the Datafusion website. Please feel free to suggest improvements, clarification points.

This is a blog post explaining the inner working of the ordering requirement analysis in the Datafusion. Actual rule and analysis is more involved than described in the blog post. However, I think the detail in the blog post is good enough for most use cases.

[alamb]

I will review this later today

[alamb]

This is great @akurmustafa -- thank you.

I have some high level feedback below. Let me know if it makes sense or if you would like me to help implementing some of it

Introduction / Setting

I think the blog would benefit from a few more examples and some more background / introductory content that connects the algorithm described in this post to something users experience. As it is currently written, I think it would be hard for people who do not already have a good understanding of the concepts here to understand the rest of the post.

I left some specific comments

Formatting

I rendered this locally using the instructions here: https://github.com/apache/datafusion-site?tab=readme-ov-file#setup-for-docker

Some of the formatting looks like it isn't quite working as intended:

[akurmustafa]

Thanks @alamb for your feedback as always. I will try to adress those points. Please feel free to add additional feedback as we revise through this work. I am happy to address them to improve readability and flow of the document.

[akurmustafa]

Hi @alamb, I have addressed the points you mention. I also changed the order of some sections to make the post more clear. I am happy to address reviews by the community.

[ozankabak]

LGTM - good content for the DF blog

[alamb]

Giving it another read now

[alamb]

Thank you so much @akurmustafa -- this is great! I went through the text quite carefully and have several suggestions, though in my opinion none are required.

If you choose to make any I will be happy to go through and do a final proof reading round as well.

FYI @wiedld and @Omega359 as I think both of you have looked at the code that implements the algorithm described in this post

[akurmustafa]

Thanks @alamb for the detailed feedback. I have incorporated your suggestions. Thanks @Omega359 also for catching the typos, I have fixed them in the commit.
I did a proof read to see whether formatting, rendering, links are OK? However, second look is always welcome.

[alamb]

Thank you so much @akurmustafa and @ozankabak -- I pushed two commits with minor changes (spelling/code formatting) and then to update the date to today.

I think this is ready to publish!

[Omega359]

Minor grammatical updates, one typo.

[akurmustafa]

Minor grammatical updates, one typo.

Thanks @Omega359, for such a detailed read-through. I have added those changes in the commit

[alamb]

Thanks again @akurmustafa @Omega359 and @ozankabak -- I think this one looks really nice

[alamb]

It is posted to the site now: https://datafusion.apache.org/blog/2025/03/11/ordering-analysis/