You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
awesome project! I'm trying to use datafusion to query across 100s of thousands of partitioned parquet files in S3. Partitioning is e.g. experiment=A/measurement=B/date_id=20230101/file.parquet.
Now say I want to query SELECT * FROM table WHERE experiment='A' AND measurement='B' AND date_id=20230101.
Depending on what I do I get very different performance:
If I first register_listing_table with prefix experiment=A/measurement=B/ and only specify date_id in ListingOptions.with_table_partition_cols , and then send query SELECT * WHERE date_id=20230101 that's very fast.
If I just register_listing_table without prefix, and in ListingOptions.with_table_partition_cols specify experiment, measurement, date_id, and send the full query SELECT * WHERE experiment='A' AND measurement='B' AND date_id=20230101, that's very slow. (Still works!)
It makes sense to me that the first time, we have to list all files, so (2) is slower. But it's also slower on repeated queries.
In ListingTable, allow specifying the order of partitioning columns (e.g. in this case [experiment, measurement, date_id]), and if specified filters are a prefix of this, use that to only list beyond that prefix. E.g. here that'd lead to only listing files after experiment=A/measurement=B/.
Do any of these ways already exist and I just didn't find them? Or already have tickets?
To Reproduce
No response
Expected behavior
No response
Additional context
No response
The text was updated successfully, but these errors were encountered:
When discovering partitions for pruning, if we specify
no partition columns, we call `list_all_files`, which
uses the `list_files_cache` if it exists and is filled.
If we specify partition columns, before this change, we
recursively list files in the object store to discover
partitions. That happens on every request, and listing
files e.g. in AWS S3 can be slow (especially if it's
100k+).
With this change, if the `list_files_cache` exists and is
filled, we get all files from there and use that to discover
partitions.
Closesapache#9654
Describe the bug
Hey,
awesome project! I'm trying to use datafusion to query across 100s of thousands of partitioned parquet files in S3. Partitioning is e.g.
experiment=A/measurement=B/date_id=20230101/file.parquet
.Now say I want to query
SELECT * FROM table WHERE experiment='A' AND measurement='B' AND date_id=20230101
.Depending on what I do I get very different performance:
If I first
register_listing_table
with prefixexperiment=A/measurement=B/
and only specifydate_id
inListingOptions.with_table_partition_cols
, and then send querySELECT * WHERE date_id=20230101
that's very fast.If I just
register_listing_table
without prefix, and inListingOptions.with_table_partition_cols
specify experiment, measurement, date_id, and send the full querySELECT * WHERE experiment='A' AND measurement='B' AND date_id=20230101
, that's very slow. (Still works!)It makes sense to me that the first time, we have to list all files, so (2) is slower. But it's also slower on repeated queries.
Afaict the actual filtering happens in https://github.com/apache/arrow-datafusion/blob/cf0f8eececd37f593b811320f89c0edd00fd3945/datafusion/core/src/datasource/listing/helpers.rs#L328
which calls
list_partitions
.I can think of two ways to speed this up:
ListingTable
's filesListingTable
, allow specifying the order of partitioning columns (e.g. in this case[experiment, measurement, date_id]
), and if specified filters are a prefix of this, use that to only list beyond that prefix. E.g. here that'd lead to only listing files afterexperiment=A/measurement=B/
.Do any of these ways already exist and I just didn't find them? Or already have tickets?
To Reproduce
No response
Expected behavior
No response
Additional context
No response
The text was updated successfully, but these errors were encountered: