Support dynamic filter for hash join

[SilenceJD]

Support dynamic filter for hash join. Currently, Presto 0.241+ supports local dynamic filtering for broadcast inner-joins, but does not have a dynamic filtering for hash inner-joins. This PR add a new session property "ENABLE_HASH_JOIN_DYNAMIC_FILTERING" which support dynamic filter for hash join.

== RELEASE NOTES ==

General Changes

* Add `HashJoinDynamicFilterSourceOperator` to collect filter data info before hash table build in build side.

* Add `DynamicFilterService` to aggregate filter data info in coordinator and broadcast it to the probe side.

* Add `DynamicPageFilter` to do page filtering in the probe-side by filter data info which receive from the build side.

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: SilenceJD (dbf8b24)

[highker]

Haven't looked into design yet. Maybe let's fix the coding style first?

[highker]

Can we use airlift json?

[highker]

We don't use any abbreviations or acronyms in our code base. Please spell out bf. Also, there are a lot of other places in this PR having this issue. Could you go over them and fix accordingly?

[highker]

"bloom" is a weird name. Use "bloomfilter"

[SilenceJD]

Haven't looked into design yet. Maybe let's fix the coding style first?

Sure, I will fix them right away.

[rohanpednekar]

Gentle Ping! 🙏

[highker]

@kewang1024, could you help to review this PR?

[rubenssoto]

Its is a great contribution. only a gentle reminder. :)

[yingsu00]

Is this still being worked on? @SilenceJD

[rohanpednekar]

@SilenceJD Hope all is well with you. Just checking in to see if you are still planning to work on this PR?
@v-jizhang Do you think you can pick this up further to get it merged if needed? Thanks!

[SilenceJD]

Is this still being worked on? @SilenceJD

@yingsu00 yes, I'd be happy to follow up on this PR if community can give some feedback and suggestions.

[SilenceJD]

@SilenceJD Hope all is well with you. Just checking in to see if you are still planning to work on this PR? @v-jizhang Do you think you can pick this up further to get it merged if needed? Thanks!

@rohanpednekar Thank you for your concern. I can continue to work for this request, and I'm glad you and your partners can join in.

[rohanpednekar]

Thanks @SilenceJD.

@yingsu00, Do you think you can further review this PR and provide your feedback for @SilenceJD ?

[cliandy]

@SilenceJD @rohanpednekar do we have any updates on this? This would be very helpful for us and would be happy to work on this to get it out asap.

[lfyzjck]

We had the same problem, which cause prestodb had bad perf in tpcds q17 compare with trino

[rohanpednekar]

@atris Ses if you can pick this one?

[tdcmeehan]

@SilenceJD it looks like this slipped between the cracks. Nonetheless, this would be a great feature, and could significantly improve TPC-DS performance, so I'm hoping you could help us pick this back up. We would like to help you push this forward--are you prepared to address review feedback and help move it forward? If so, we can review and help.

[SilenceJD]

@SilenceJD it looks like this slipped between the cracks. Nonetheless, this would be a great feature, and could significantly improve TPC-DS performance, so I'm hoping you could help us pick this back up. We would like to help you push this forward--are you prepared to address review feedback and help move it forward? If so, we can review and help.

@tdcmeehan Thank you very much for your reply, I am glad that you can help and start this work again. But this pr is already far behind the master branch, I may need to resubmit a new pr based on the master branch and improve some of the previous implementations. Please give me some time~

[tdcmeehan]

@SilenceJD that's wonderful--we'll be ready to look once you've rebased and are ready for review. Thanks a lot!

[sajjoseph]

Is this still being working on?
@SilenceJD - hope you can bring this to a closure.
Thanks!

[tdcmeehan]

@sajjoseph I'm working on this, see prestodb/rfcs#54