[prev] 13 [next]

Clustering Index (cont)

Index on non-unique ordering attribute Ac.

Usually a sparse index; one pointer to first tuple containing value.

Assists with:

  • range queries on Ac   (find lower bound, then scan data)
  • pmr queries involving Ac   (search index for specified value)
Insertions are expensive: rearrange index file and data file.

Deletions relatively cheap (similar to primary index).

(Note: can't mark index entry for value X until all X tuples are deleted)