Skip to content

GH-36765: [Python][Dataset] Change default of pre_buffer to True for reading Parquet files #37854

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.

Sign up for GitHub

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

Merged
jorisvandenbossche merged 4 commits into from
Oct 6, 2023
Merged

GH-36765: [Python][Dataset] Change default of pre_buffer to True for reading Parquet files #37854

Show file tree
Hide file tree
Changes from 1 commit
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
16e56fa
[Python][Dataset] Change default of pre_buffer to True for reading Pa…
jorisvandenbossche Sep 25, 2023
ed1b46f
Merge remote-tracking branch 'upstream/main' into gh-36765-parquet-pr…
jorisvandenbossche Oct 5, 2023
0c21100
update defaults on C++ as well
jorisvandenbossche Oct 5, 2023
f6e56e1
update docstring
jorisvandenbossche Oct 5, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
4 changes: 2 additions & 2 deletions python/pyarrow/_dataset_parquet.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -666,7 +666,7 @@ cdef class ParquetFragmentScanOptions(FragmentScanOptions):
Disabled by default.
buffer_size : int, default 8192
Size of buffered stream, if enabled. Default is 8KB.
pre_buffer : bool, default False
pre_buffer : bool, default True
If enabled, pre-buffer the raw Parquet data instead of issuing one
read per column chunk. This can improve performance on high-latency
filesystems.
Copy link
Member

@lidavidm lidavidm Oct 5, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Possibly we should improve the docstring to also mention that you should disable this if you are concerned with memory usage over throughput? (Also, possibly make it clear that "high-latency filesystems" is likely to mean object stores like S3, GCS, etc.)

Copy link
Member Author

@jorisvandenbossche jorisvandenbossche Oct 5, 2023

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yep, will do

Expand All @@ -688,7 +688,7 @@ cdef class ParquetFragmentScanOptions(FragmentScanOptions):

def __init__(self, *, bint use_buffered_stream=False,
buffer_size=8192,
bint pre_buffer=False,
bint pre_buffer=True,
thrift_string_size_limit=None,
thrift_container_size_limit=None):
self.init(shared_ptr[CFragmentScanOptions](
Expand Down
10 changes: 5 additions & 5 deletions python/pyarrow/tests/test_dataset.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -784,28 +784,28 @@ def test_parquet_scan_options():
opts2 = ds.ParquetFragmentScanOptions(buffer_size=4096)
opts3 = ds.ParquetFragmentScanOptions(
buffer_size=2**13, use_buffered_stream=True)
opts4 = ds.ParquetFragmentScanOptions(buffer_size=2**13, pre_buffer=True)
opts4 = ds.ParquetFragmentScanOptions(buffer_size=2**13, pre_buffer=False)
opts5 = ds.ParquetFragmentScanOptions(
thrift_string_size_limit=123456,
thrift_container_size_limit=987654,)

assert opts1.use_buffered_stream is False
assert opts1.buffer_size == 2**13
assert opts1.pre_buffer is False
assert opts1.pre_buffer is True
assert opts1.thrift_string_size_limit == 100_000_000 # default in C++
assert opts1.thrift_container_size_limit == 1_000_000 # default in C++

assert opts2.use_buffered_stream is False
assert opts2.buffer_size == 2**12
assert opts2.pre_buffer is False
assert opts2.pre_buffer is True

assert opts3.use_buffered_stream is True
assert opts3.buffer_size == 2**13
assert opts3.pre_buffer is False
assert opts3.pre_buffer is True

assert opts4.use_buffered_stream is False
assert opts4.buffer_size == 2**13
assert opts4.pre_buffer is True
assert opts4.pre_buffer is False

assert opts5.thrift_string_size_limit == 123456
assert opts5.thrift_container_size_limit == 987654
Expand Down