Today I decided to take xVelocity and columnstore indexes for a spin and see whether this technology lives up to all the hype. I tested it on a 2Gb fact table with about 21M rows. First I wrote a query that joined the fact table with a Customer dimension and returned the total revenue grouped by a customer occupation. On my laptop equipped with a solid state hard drive, the query consistently run in about 10 seconds. After that I created a Nonclustered Columnstore Index on the fact table (took about 2 minutes to complete with a size of the index at about 10% of the data size). After the columnstore index was created, the query ran under a second every single time. I can definitely see that 10x-100x increase in query performance that is advertised by Microsoft. I am sure that on a larger data set I would probably need to spend a little bit more time tuning my columnstore index performance, but I have to say I was really impressed with the dramatic increase in performance delivered by xVelocity and columnstore indexes.