I'm watching Paul S Randal's index fragmentation video at here. In the video, Paul said that GUID or other random values as high-order key will cause page splits which in turn will cause index fragmentation
. The reason is that, random values means the new records will be inserted into random pages, which will probably cause page split on random page. My question: if the new records are not inserted into random pages, page split still can occur, right? If this is true, what's the difference in between the two cases?
BTW, does high-order key mean primary key
? I'm not a native English speaker and not sure of it.
Thanks.