Clustered Indexing
Clustered indexing can be both dense and sparse. Let's see how
Quick recap: clustered indexing is done on non key attributes and it is sorted according to that attribute. The idea behind clustered indexing is to have one entry per unique value, as once we have a pointer to that location we can find other entry for the value as it is sorted.
Case 1: non key attribute has a unique value
It will be dense indexing as we have to give every value entry in the index table
Case 2: non key attributes isn't unique
In this case, we can have just one entry for every unique value in the index table.
Secondary Indexing
Secondary indices must be dense, with an index entry for every search-key value, and a pointer to every record in the file.
In primary indexing, its main file is sorted according to candidate keys (making values unique) so we can simply have a pointer one entry per block in the main file. We can simply go to that block and get the required data. We don't have this luxury in secondary indexing as the main file is unsorted and we may have done sorting according to key or non key attribute.
Source: http://www.cs.sfu.ca/CourseCentral/354/zaiane/material/notes/Chapter11/node8.html
