Add aggreate key dependency optimizer #1517
Merged
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Outdated suggestions cannot be applied.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.
This PR adds AggregateKeyDependencyOptimizer which analyzes user input group by keys' dependency and split into
keys
anddependentKeys
. AdependentKey
has functional dependency on akey
.keys
will be hashed during hash aggregate whiledependentKeys
will be directly materialized into f-table.Consider the following example
Based on user input, both
a.ID
anda.age
are stated as hash keys. However, sincea.ID
is the primary key,a.age
has functional dependency ona.ID
so we only need to hasha.ID
in hash aggregate.This optimizer is more important given Cypher grammar because the grammar itself doesn't differentiate
SELECT
andGROUP BY
which means user cannot group by a subset of selection list. Another case is Cypher allows group byNode
orRel
which by definition is to group by all properties ofNode
orRel
. A more reasonable solution is to only group by their internal IDs and treat properties as dependentKeys.Performance benchmark
Dataset: LDBC100
Machine: M1 Pro, 16GB memory, maximum 4 threads
Query:
MATCH (a:Person)-[:knows]->(b:Person) RETURN keys, COUNT(*) LIMIT 1;
(Note that Aggregate is a Sink operator that is not affected by LIMIT, LIMIT 1 is to reduce print size).