Data skipping for Hudi connector

[codope]

Description

Use Hudi column stats to skip data and improve query latency.
Stacked on top of #16034

Additional context and related issues

Release notes

( ) This is not user-visible or docs only and no release notes are required.
( ) Release notes are required, please propose a release note for me.
(x) Release notes are required, with the following suggested text:

# Section
* Support for data skipping with column statistics in Hudi connector.

[tooptoop4]

any benchmark?

[codope]

@tooptoop4 We benchmarked with 6-month GitHub archive data (220 GB, 450 million records) and observed significant improvements with column stats and clustering enabled.

The design is almost similar to prestodb/presto#18606 and @xiarixiaoyao also benchmarked with a bigger dataset - SSB (1.5 TB, 12 billion records) - and the results were quite impressive as you can see in that PR.

[dertodestod]

First of all a big thanks from me for the work on the connector and on hudi in general 👍. Is there a rough ETA on when this will be merged and be available in trino? Thanks again.

[mosabua]

@codope do you plan to rebase and update this PR soon?

[songpcmusic]

Is there any plan to support the use case where an operation such as element_at(map_column, 'key') =|>|< 'value' is required?

(Note: The =|>|< symbol is not a standard operator in most query languages, so for an accurate translation, it would be helpful to provide additional context or clarification on what the intended operation is.)

[codope]

Is there any plan to support the use case where an operation such as element_at(map_column, 'key') =|>|< 'value' is required?

(Note: The =|>|< symbol is not a standard operator in most query languages, so for an accurate translation, it would be helpful to provide additional context or clarification on what the intended operation is.)

The current PR does not support that. In Hudi 1.0, we are adding functinal indexes which can support skipping data based on function/expression on column(s).

[github-actions]

This pull request has gone a while without any activity. Tagging the Trino developer relations team: @bitsondatadev @colebow @mosabua

[mosabua]

👋 @codope .. I assume you are continuing this work at some stage and will leave the PR open.

[codope]

I am closing this PR. Once we upgrade the Hudi version in Trino (with the updated Hadoop-independent abstraction), we'll revive or create a new PR for data skipping.