I have 2 tables which are joined. All entries in both tables are unique. Now in the query planner and exec summary it shows more records coming out of the hash join operator than both input sides combined.
Left table: 589,740 records Right table: 116,776 records Intermediate join result: 92,291,634 records
There is a group by afterwards which brings the result back down to the left table number of records: Eventual end result: 589,740
My assumption was that the join result can be only the number of records of either the left or the right side (left side in this case as it is a left join) but not more without having duplicate records.