1. Add a clustered index to each table
2. Keep indexes narrow, especially the clustered index
3. Make sure that the columns in the clustered index aren't updated
4. Make clustered indexes monotonically increasing whenever possible
5. Design the column order of indexes around how it will be used
6. Make and use covering indexes where possible
>4. Make clustered indexes monotonically increasing whenever possible
Thought I would double check, I'm assuming this clustered index will be used by the application to query data in DB?
Or is the clustered index column added in part to make a record have a unique key(per recommended practices)?
If it's the second one (auto-increment) I believe we could improve utilization of clustered index by:
Picking a unique column (e.g. SSN though not 100% unique)
Picking potentially even a DATE column, if this table might be used to query a lot of date ranges. This could save some IO.
Though as always "it depends" :hehe:
--------------------------------------------------
...0.05 points per day since registration... slowly crawl up to 1 pt per day hopefully 😀