90% of the data errors I encounter in our Elasticubes are due to differences in capitalization within joined text fields. Most versions of SQL I have worked with ignore case when joining text unless you specify a certain collation. I have wasted a lot of time investigating bad data, specifying Upper() (which can cause performance issues in a large data set), and making calculated columns to work around case sensitivity.
Question: Does anybody actually prefer that all text matching considers case? If so, why? I have searched the internet and spoken to many people to compile reasons where that would be useful and only found a small number of edge cases.
Does anyone else encounter data errors or issues due to this frequently? Are you going to check your cubes now? Hahaha.
If you feel strongly either way, please post here. I am trying to make a case against case sensitivity for future releases.
Please sign in to leave a comment.