medplum / medplum / 27441976931

92%

main: 92%

push

github

feat(data warehouse): restore ORDER BY (#9435)

ORDER BY in the query makes sure the data is ordered in the Parquet files,
which means the metadata in the Parquet files will contain the proper metadata
for each row group, such as min/max.

This metadata is then used by Iceberg and query engines, and suboptimal metadata
will result can result in slow/inefficient query plans.

The risk of spilling to disk is slow, as the data is small and ECS Fargate ephemeral disk
tasks have a default of 20GB, which is more than enough for each resource type.

Signed-off-by: Karl Pietrzak <karl@medplum.com>

24568 of 27360 branches covered (89.8%)

Branch coverage included in aggregate %.

36835 of 39031 relevant lines covered (94.37%)

23235.3 hits per line