Scoutdb aggregation and filter pushdown optimizations - #1
Draft
bendischinger wants to merge 13 commits into
Draft
Conversation
This commit enhances the pushdown optimization logic to better handle filter conditions and improve query performance.
Add support for pushing down aggregation operations to storage engines for improved query performance. This includes partition-aware aggregation handling and integration with the query planner.
Add a max_join_build_rows session variable (Uint64, dynamic, default 0=unlimited) and instrument hashLookupGeneratingIter to abort with ErrHashLookupBuildExceededMemory when the build side exceeds the cap. Also OR-in the MemoryManager HasAvailable backstop so a runaway join returns an error instead of OOMing the process.
partitionGroupedAggregationIter.computeAll accumulates all merged groups into an unbounded map. Check ctx.Memory.HasAvailable() each time a new group is added and return ErrNoMemoryAvailable when the MemoryManager reports it is exhausted, so a high-cardinality group-by returns an error instead of OOMing the process. Mirrors the hash-join build-side backstop.
Allow COUNT/SUM/MIN/MAX/AVG over a scalar expression argument (e.g. SUM(CASE WHEN size < 1024 THEN 1 ELSE 0 END)) to be pushed down to AggregableTable integrators, for both ungrouped and grouped (GROUP BY) queries, not just bare columns. - Add optional Expr field to sql.ColumnAggregation - Analyzer: extract expression aggregations for the ungrouped, COUNT, and grouped paths; require the expr to reference at least one table column and reject nested aggregates/subqueries/aliases (SUM/AVG still require a safe numeric type). COUNT(*)/COUNT(1) keep their existing bare-count representation - PartitionAggregation/PartitionGroupedAggregation: resolve type/display from Expr when present - memory table: evaluate Expr per row, remapping GetField indices to the full schema, for both ungrouped and grouped compute paths - Test ungrouped, COUNT, and grouped conditional aggregation pushdown across multiple partitions
Introduce a storage-driven join pushdown feature that lets a table evaluate an INNER equi-join against a co-partitioned table at the partition level, mirroring the existing aggregation pushdown. - sql: add JoinableTable interface + JoinPushdown spec (tables.go), PartitionJoin plan node, and its rowexec builder - analyzer: add pushdownPartitionJoins rule (runs before join physicalization) that collapses Filter-over-InnerJoin and bare InnerJoin into a PartitionJoin, pushing qualifying WHERE predicates into the table; unwrap PartitionJoin in getResolvedTableForAgg so aggregation composes over a pushed-down join - memory: implement JoinableTable (opt-in, off by default) with a partition-level hash join; store the opt-in flag in TableData so it survives session table reconstruction - plan printer: show On(...), Right(...), and pushed Filters(...) - docs: add JOINPUSHDOWN.md, cross-reference from AGGPUSHDOWN.md - tests: TestPartitionJoinPushdown and TestPartitionJoinAggregationComposition
This file contains hidden or 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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
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.
Just putting this out here to get some eyes on it for now.