Just a side not for anyone that is interested.

I've created a table on my test system with two columsn

VARCHAR(900) --Max size for a clustered index

VARCHAR(32)

even when I create two index one clustered on the first col and one nonclustered on the second column, my query always uses parallelism and always uses the first index(the bigger one) and always runs slow.

I'm going to trying to find out how the optimizer decides which index/column etc to use for a count 🙂

wish me luck