Adaptive Server runs a query in serial when:
There is not enough data to benefit from parallel access.
The query contains no equijoin predicates like:
select * from RA2, RB2 where a1 > b1
There are not enough resources, such as thread or memory, to run a query in parallel.
Uses a covered scan of a global nonclustered index.
Tables and indexes are accessed inside a nested subquery that cannot be flattened.
