It would be great if we pushed down the count to each partition and aggregated the aggregates, even if there is a where clause.
From the forums:
A simple count of the key_value table in the key_value example is giving the error:
Output from SQL stmt overflowed output/network buffer of 10mb. Try a "limit" clause or a stronger predicate.
This is through a stored procedure.
I know it is multi-partitioned query and performance will be impacted but surprised to see it hitting the buffer limits. Any work arounds to get to count of huge tables ?