Oracle Internal Arch. (contd.)
Cluster : Data from one or more (master-detail) tables is kept in one segment based on some field(s). The cluster key is a primary key in one segment, a foreign key in another segment, and so on.
- hash cluster : Rows are organized using a hash algorithm on cluster key. Rows with same hash values are clustered nearby. This is good in case of queries that specify equality but not for range queries.
- Full table scans suffer in clusters. Clusters are rarely used in data warehousing. Even in OLTP it may not be very efficient.