• rice.tx (2/21/2013)


    We have a very similar situation, and are looking forward to upgrading and making use of filtered indexes.

    It's not clear to me... did you try using filtered indexes with just the "status" field, and still had problems with complex queries? Or were the problems you experienced in prior versions, where the "status" index was a regular (non-filtered) index.

    Thanks.

    Hi there, originally the status field was a normal index, and then I started experimenting with a filtered index and a few covering fields, and eventually came to the conclusion that adding all the columns would be highly beneficial in this scenario. This was due to the steady increase in disk IO over the years starting to create minor performance issues. Although I show a very simple example the real world queries are actually very complex but the common theme was very frequently that I was looking to take all the ACTIVE customer orders and look at the ACTIVE order routing records for those orders so if I had 'tables' of just the active orders the query optimiser would have a much simpler time without thinking about bookmark lookups etc. I hope this helps.