You know how if you drop a clustered index (turning the base table into a heap), SQL Server reuses the data pages and just drops the b-tree levels above the leaf? @dbBerater wrote all about it in social.technet.microsoft.com/wiki/contents/… Anyway, it turns out there's an exception. If Accelerated Database Recovery is enabled, SQL Server doesn't reuse the data pages and writes new ones instead. (To be fair to Uwe, that feature wasn't a thing when that article was written). There's probably an in-row diff/persistent version store/sLog reason for the difference, but I don't know what it is. @pskountrianos would know, I'm sure. There might be other exceptions where the keeping the same data pages optimization can't be applied as well. You can never really be sure of anything with SQL Server 😀 Curiously, even when ADR prevents the clustered data pages being reused, the newly allocated pages have previous- and next-page pointers just like the original clustered index.