Question

What are the benefits/drawbacks of using a case insensitive collation in SQL Server (in terms of query performance)?

I have a database that is currently using a case-insensitive collation, and I don’t really like it. I would very much like to change it to case sensitive. What should I be aware of when changing the collation?

Answer 1

(I added this as a separate answer because its substantially different than my first.)
Ok, found some actual documentation. This MS KB article says that there are performance differences between different collations, but not where you think. The difference is between SQL collations (backward compatible, but not unicode aware) and Windows collations (unicode aware):

Generally, the degree of performance difference between the Windows and the SQL collations will not be significant. The difference only appears if a workload is CPU-bound, rather than being constrained by I/O or by network speed, and most of this CPU burden is caused by the overhead of string manipulation or comparisons performed in SQL Server.

Both SQL and Windows collations have case sensitive and case insensitive versions, so it sounds like that isn’t the primary concern.

Another good story “from the trenches” in Dan’s excellent article titled “Collation Hell“:

I inherited a mixed collation environment with more collations than I can count on one hand. The different collations require workarounds to avoid “cannot resolve collation conflict” errors and those workarounds kill performance due to non-sargable expressions. Dealing with mixed collations is a real pain so I strongly recommend you standardize on a single collation and deviate only after careful forethought.

He concludes:

I personally don’t think performance should even be considered in choosing the proper collation. One of the reasons I’m living in collation hell is that my predecessors chose binary collations to eke out every bit of performance for our highly transactional OLTP systems. With the sole exception of a leading wildcard table scan search, I’ve found no measurable performance difference with our different collations. The real key to performance is query and index tuning rather than collation. If performance is important to you, I recommend you perform a performance test with your actual application queries before you choose a collation on based on performance expectations.

Hope this helps.