Table and Index Organization - Relational Database Index Design and the Optimizers - page 27

A table can have only one clustering index but, at a given point in time, several
indexes may be clustered. The CLUSTERRATIO of an index is a measure of the
correlation between the order of the index rows and the table rows. It is used by
the optimizer to estimate I/O times.
DB2 tables normally have a clustering index.
In SQL Server, the index that stores the table rows is called clustered ;a
clustered index is only defined if an index-only table is required. The other
indexes (SQL Server term: nonclustered indexes ) point to the clustered index .
In Oracle, the word cluster is used for the option to interleave table rows
(clustered tables). It has nothing to do with the clustering index that we have
taken to determine the sequence of the table rows.
DB2 for LUW V8 has an option called multidimensional clustering ;this
enables related rows to be grouped together. Refer to Chapter 13 for more details.
Important
In the diagrams throughout this topic, C is used to mark the index that defines the
home page for a table row that is being inserted . In our calculations, the table rows
are assumed to be in that same order . For a product that does not support a clustering
index in this sense, the order of the table rows is determined when reorganizing and
reloading the table.