Core: Support aggregated basic stats in partition summary

[deniskuzZ]

ref: apache/hive#5498

[pvary]

@deniskuzZ: Could you please provide a short description what data is stored in the summary and in what format?

I think it is important to understand the cost for keeping this stat up-to-date. How costly is to calculate it, and what is the data size increase caused by this change.

@findepi: Could this be useful for Trino? Does Trino have some optimiyation like this?

[pvary]

This discussion could be relevant here too: https://lists.apache.org/thread/0q1csnkfg8jc11zo1dlssjkr4v8s8zz0

[deniskuzZ]

@pvary, unfortunately, that won't work. I was looking for an easy way to get basic partition stats, however, I missed the part that iceberg only keeps the changed partitions in a SnapshotSummary. Aggregation with just the prev snapshot value is not enough, it requires loop through all the snapshots.

table.newFastAppend().appendFile(FILE_A).commit();
partitions.data_bucket=0 -> added-data-files=1,added-records=1,added-files-size=10,total-records=3,total-files-size=30,total-data-files=3,total-delete-files=0,total-position-deletes=0,total-equality-deletes=0

table.newFastAppend().appendFile(FILE_B).commit();
partitions.data_bucket=1 -> added-data-files=1,added-records=1,added-files-size=10,total-records=2,total-files-size=20,total-data-files=2,total-delete-files=0,total-position-deletes=0,total-equality-deletes=0

table.newFastAppend().appendFile(FILE_A).commit();
partitions.data_bucket=0 -> added-data-files=1,added-records=1,added-files-size=10,total-records=3,total-files-size=30,total-data-files=3,total-delete-files=0,total-position-deletes=0,total-equality-deletes=0

do you think it's worth doing it in SnapshotSummary or is there some simpler/better way like create or update the partition stats puffin file right after the commit?

[deniskuzZ]

Found partition stats tracker issue #8450 with the following design doc: https://docs.google.com/document/d/1vaufuD47kMijz97LxM67X8OX-W2Wq7nmlz3jRo8J5Qk
But it doesn't seem to be completed yet: #11216

[pvary]

And here is the relevant mailing list thread: https://lists.apache.org/thread/knl1ol7s1o2p7rglgl2mm8c5mc2pk0sx

@ajantha-bhat: Are you still working on the proposal?