How the SQL compiler works with range-clustered tables

The SQL compiler handles the range-clustered table (RCT) in a similar way to a regular table that has a secondary B+ tree index. Rather than working through a B+ tree index to determine the record's location or record identifier (RID), RCT uses a functional lookup involving the record key values and the algorithm from the range definition. This is similar to having an index because a key value is used to obtain the RID quickly.

When working to determine the best access path to required data, the SQL compiler uses statistical information kept about the tables. Index statistics are collected during a table scan when a RUNSTATS command is issued. For an RCT, the table is modeled as a regular table, and the index is modeled as a function-based index.

Order of records in the table is not guaranteed when creating a range-clustered table allowing overflow.