Are Filtered Indexes Really that Bad?
Posted on 2014-09-21
I have a what I believe is a strong case for using a filtered index. Based on a simple select statement there would be a 50% savings in the number of rows queried/stored. DBCC SHOW_STATISTICS makes a much stronger case. The table has 5 steps in the histogram and based on EQ_ROWS the savings might be as much as 80%. The statistics are current.
There are a lot of caveats with the SET options I am reading and since the entire code base of the application is not available I feel like the possibility of failed inserts is risky with practically no knowledge of the clients connectivity options.
I am interested in knowing whether there is an easy or even worthwhile method to determine the connection SET options for a particular connection.