I’m am rediscovering many features of the DOL data page and index page structures.
1. On non-leaf DOL index pages, only the page # and shortest distinguishable part of the key is written.
If “Carl” was the last key on index leaf #2 and
“Connie” was the first key on index leaf #3
then the non-leaf page would hold “Co”,3 because “Co” was enough to distinguish Connie from Carl…
This method possibly allows the non-leaf pages to maintain more entries than APL non-leaf pages.
2. On leaf DOL pages, the whole key is stored so that the index can retain the covered read features.
3. Leaf index pages store duplicate key values once per page with multiple rids.
4. Leaf index pages has a sorted offset table in the page; key+rids pairs are not sorts.
5. DOL Pages do not use the Address Lock Spinlock (added 7/28/2014)