Presto performance with delta table.

[anand-sharma-29]

Presto with delta table performance issue :

We have two version of a delta table with hourly partition.

1. With ~2000 partitions (based on date and hour)
2. With ~300 partitions (based on date and hour, but less back dated data available in 2 then 1)

Query which targets same hour level partition, gives better performance on (2) compared to (1). Example query - Select id from table where date = ? and hour = ? (both table partitions have equal no of file and data)

Does number of partitions and hence no of manifest files of a delta table impacts query performance. How can we improve on table 1 ?

We are using - Athena + Glue combination