Restructure Your Data Table for Improved Compression
I was relaxing during my vacation thinking some more about column compression in Power Pivot. One of the main things to know about compression is that a high level of cardinality is your main enemy (ie a large number of unique values will mean poor compression). I started to think about how I could reduce the cardinality of one or more columns in a large data table I use for a client. This blog covers the process I went through to test a couple of concepts – warts and all – including a simple error I made during the early testing for this blog post. I think you can sometimes learn more when you make and then find a mistake, hence I have kept the error I made in this post for others to see along with the fix. What I think is interesting about this is the process of discovery to find the problem.