- How much oxygen and hydrogen are in the soil of Mars?
- Why is hydrazine used in arcjet thrusters?
- How does Accion System's TILE propulsion module compare to an equivalently sized Hall effect or an Ion thruster?
- Particles changing color based on camera position?
- Blender .STL exports are too small when imported in Sli3er
- Version 2.8 or 2.79 to start learning Blender?
- Looking for advice on realizing Blender scanner type project, maybe using sensors in Game Engine?
- Whether every router in a AS must have its RouterID?
- What does “pacing his study” mean in this context?
- “We do me old” What does it mean?
- What is it called when someone says “I guess I expected too much from you”
- Difference was killed and killed
- Magento 2 Add New Integration error
- Magento 2 How to make changes to Bundle product edit detail page?
- How to specific pass fields in custom rest api for get product Information
- Magento2.1 Consumer is not authorized to access %resources
- Set Category wise Discount in Magento2
- Import product CSV to magento
- Future Lucy's right hand?
- Nanatsu no Taizai: Whats the first song at the beginning of episode 1 OVA
Newbie Clustered Index question - when to NUSE?
I have a bit of a newbie question, which I feel I should really know having dealt with databases for so long, but...
I know the general advice for indexes is NUSE, but a lot of my tables have a non-unique column that is a not-quite-foreign-key from another table, and will be 90% of the predicate key for any queries. The ID will never be used (because of awesome design going back years).
I've attached an example, with horrible anonymising.
So, my question is, if the ...eID is not Unique, and not Ever-Increasing, would I be better off moving my Clustered Index to a PK that is never used for anything other than identity/index, or would I be better using a combo of my ...eID/PK for a Unique Clustered Index (replacing the uniquifier with my actual pk int)?
As you can see, there has been multiple NCIX made on the one column I'd be proposing, and aggregating them, and leaving some space in my FILLFACTOR sounds reasonable to me, which might lead to higher IO, but I'm unsure