Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

GH-43695: [C++][Parquet] flatbuffers metadata experiments #43793

Open
wants to merge 9 commits into
base: main
Choose a base branch
from

Conversation

alkis
Copy link
Contributor

@alkis alkis commented Aug 22, 2024

This is an annotated attempt to use flatbuffers as metadata for parquet. The goals are:

  1. flatbuffers "parse" extremely fast compared to thrift which
    • cuts down on critical path latency of processing parquet files
    • is so fast, O(n) effects of "parsing" metadata vs scanning 1 column are eliminated
  2. flatbuffers are typically bulkier than thrift, in this PR there are a multitude of optimizations to shrink the size of flatbuffer metadata
  3. keep the flatbuffer object model similar to that of thrift to facilitate easier migration to new metadata format

To run experiments:

mkdir arrow/src/o
cd arrow/src
cmake o --preset ninja-benchmarks
ninja -Co && o/relwithdebinfo/parquet-metadata3-benchmark path-to-footers/*

@alkis alkis requested a review from wgtmac as a code owner August 22, 2024 17:21
Copy link

Thanks for opening a pull request!

If this is not a minor PR. Could you open an issue for this pull request on GitHub? https://github.com/apache/arrow/issues/new/choose

Opening GitHub issues ahead of time contributes to the Openness of the Apache Arrow project.

Then could you also rename the pull request title in the following format?

GH-${GITHUB_ISSUE_ID}: [${COMPONENT}] ${SUMMARY}

or

MINOR: [${COMPONENT}] ${SUMMARY}

In the case of PARQUET issues on JIRA the title also supports:

PARQUET-${JIRA_ISSUE_ID}: [${COMPONENT}] ${SUMMARY}

See also:

@corwinjoy
Copy link

Thanks for posting this!
A few basic initial comments / questions:
Running the build, I think you want the following for instructions:

cd arrow/cpp
mkdir o
cd o
cmake .. --preset ninja-benchmarks
cd ..
ninja -Co
o/relwithdebinfo/parquet-metadata3-benchmark path-to-footers/*

Here I split the build and run. For now, it will not run until you generate footers per (https://github.com/apache/parquet-benchmark/pull/1/files).

@corwinjoy
Copy link

As part of running this code, I put in a PR with synthetic footers to the parquet-benchmark repository that can be used to get started (when setting your path-to-footers/directory).
apache/parquet-benchmark#2

@alkis alkis changed the title flatbuffers metadata experiments [GH-43695] flatbuffers metadata experiments Aug 23, 2024
@alkis alkis changed the title [GH-43695] flatbuffers metadata experiments GH-43695: flatbuffers metadata experiments Aug 23, 2024
Copy link

⚠️ GitHub issue #43695 has been automatically assigned in GitHub to PR creator.

@github-actions github-actions bot added awaiting committer review Awaiting committer review and removed awaiting review Awaiting review labels Aug 23, 2024
@kou kou changed the title GH-43695: flatbuffers metadata experiments GH-43695: [C++][Parquet] flatbuffers metadata experiments Aug 23, 2024
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

Successfully merging this pull request may close these issues.

2 participants