perf(parquet): splice buffered pages with write_all instead of io::copy (adapts #10052)#10353
Open
adriangb wants to merge 2 commits into
Open
perf(parquet): splice buffered pages with write_all instead of io::copy (adapts #10052)#10353adriangb wants to merge 2 commits into
write_all instead of io::copy (adapts #10052)#10353adriangb wants to merge 2 commits into
Conversation
Adapts the idea from apache#10052 to the PageStore splice path: drain each page blob from the store and write it straight to the output with a single write_all, instead of pumping every byte through std::io::copy's fixed 8 KiB buffer. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sd8JjF3qQMkn7htofu8bF3
Contributor
Author
|
run benchmark arrow_writer |
Contributor
Author
|
run benchmark parquet_round_trip |
|
🤖 Arrow criterion benchmark running (GKE) | trigger CPU Details (lscpu)Comparing append-column-from-pages (f77bf2f) to ee30b61 (merge-base) diff File an issue against this benchmark runner |
|
🤖 Arrow criterion benchmark running (GKE) | trigger CPU Details (lscpu)Comparing append-column-from-pages (f77bf2f) to ee30b61 (merge-base) diff File an issue against this benchmark runner |
|
🤖 Arrow criterion benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
|
🤖 Arrow criterion benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Contributor
Author
|
run benchmark arrow_writer |
|
🤖 Arrow criterion benchmark running (GKE) | trigger CPU Details (lscpu)Comparing append-column-from-pages (f77bf2f) to ee30b61 (merge-base) diff File an issue against this benchmark runner |
|
🤖 Arrow criterion benchmark completed (GKE) | trigger Instance: CPU Details (lscpu)Details
Resource Usagebase (merge-base)
branch
File an issue against this benchmark runner |
Its only caller is the arrow writer; without the feature the parquet_derive clippy job (-D warnings) flags it as dead code. Co-Authored-By: Claude Fable 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_01Sd8JjF3qQMkn7htofu8bF3
Jefffrey
approved these changes
Jul 16, 2026
Contributor
Author
|
Thanks fore the quick review! I can't merge so I'll wait for you or Andrew to do it. |
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
3 participants
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?
SerializedRowGroupWriter#10052 (by @alamb) to the currentPageStore-based writer. That PR predates Pluggable page spilling API for the Parquet ArrowWriter (PageStore) #10020 / [Parquet] route dictionary page through the PageStore #10142 and no longer applies cleanly; this PR carries its idea forward. Closes Avoid an extra copy inSerializedRowGroupWriter#10052's use case.Rationale for this change
When the
ArrowWriterflushes a row group, each column chunk's buffered pages are spliced into the output throughStreamingColumnChunkReader(aRead) driven bystd::io::copy. Every byte is copied twice — once intoio::copy's fixed 8 KiB buffer, then again intoTrackedWrite'sBufWriter— in ~len / 8 KiBwrite calls.The page blobs are already contiguous owned
Bytessitting in thePageStore; we can write each one straight to the sink with a singlewrite_all(large writes bypass theBufWriterbuffer entirely).#10052 did exactly this, but its
IntoIterator<Item = Bytes>signature can't express the fallible, one-page-at-a-time drain of a (possibly spilling)PageStore, and it assumed the pre-#10020Vec<Bytes>chunk representation.What changes are included in this PR?
StreamingColumnChunkReader(Readimpl) becomesStreamingColumnChunkPages, anIterator<Item = Result<Bytes>>thattake()s each page out of thePageStoreas it is consumed — preserving the one-page-in-memory bound for spilling backends.pub(crate) SerializedRowGroupWriter::append_column_from_pages, which writes each page blob with a singlewrite_all.begin_appended_column/finish_appended_column(same refactor as Avoid an extra copy inSerializedRowGroupWriter#10052), also used by the existingRead-based path that still backs the publicappend_column.Are these changes tested?
Covered by existing tests. Also verified that the output files are byte-for-byte identical to main for multi-row-group writes with int / string / dictionary columns under both default and small-page properties.
Local benchmark results (
arrow_writerbench, Apple Silicon), reproducing the wins from #10052's benchmark run without itslarge_string_non_nullregression:string_dictionary/defaultlist_primitive/defaultlarge_string_non_null/defaultAre there any user-facing changes?
No.
append_column_from_pagesis keptpub(crate)for now; it could be made public in a follow-up if theResult<Bytes>iterator contract is deemed a good public API (#10052 proposed a public variant).🤖 Generated with Claude Code
https://claude.ai/code/session_01Sd8JjF3qQMkn7htofu8bF3