Ballista: Implement scalable distributed joins #634
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
andygrove
force-pushed
the
ballista-scalable-join
branch
from
f597c0c
to
8acdd12
Compare
andygrove
force-pushed
the
ballista-scalable-join
branch
from
8acdd12
to
6f4cfd8
Compare
andygrove
changed the title
|
@edrevo fyi |
Dandandan
reviewed
Jul 3, 2021
| .with_repartition_joins(false) | ||
| .with_repartition_aggregations(false) | ||
| .with_physical_optimizer_rules(rules); | ||
| let config = ExecutionConfig::new().with_concurrency(2); // TODO: this is hack to enable partitioned joins |
There was a problem hiding this comment.
What is the idea here for later? I guess the repartitioning needs to be applied with concurrency=1 too to avoid inefficient plans?
There was a problem hiding this comment.
I filed https://github.com/apache/arrow-datafusion/issues/661 to discuss this
jorgecarleitao
approved these changes
Jul 3, 2021
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Which issue does this PR close?
Closes #63.
This PR removes previous hacks around partitioning and now faithfully translates the DataFusion query plan, including
RepartitionExec. I have tested with TPC-H query 12 and see consistent results between DataFusion and Ballista with the 100GB data set, where each table has 8 partitions. I have tested with multiple executors as well as single executors.There is more work to do but I think this is at a good point to merge since it fixes some correctness issues.
Rationale for this change
Ballista cannot scale well without this because work is duplicated across all partitions to load the entire left side of the join into memory currently.
What changes are included in this PR?
RepartitionExecin Ballista query plans and translate them to shufflesAre there any user-facing changes?
Query plans will change.