How to implement "Filtered Indexes" in Oracle?

Question

I know that function-based index can (kind-of) create index for only a portion of all the rows in a table,
create index idx_abc_01 on t_abc (case when end_dt > '01-APR-2008' then end_dt else null end);
/* to use the index, you have to use the same express in the filter clause */
select * from t_abc where xyz in ('O', 'S', 'X') and
     case when end_dt > '01-APR-2008' then end_dt else null end
          between '12-JUL-2008' and '15-JUL-2008';But this approach is very inflexible because the expression in the where clause has to exactly match the index creation, otherwise that filtered index will not be used.
MS SQL Server 2008 has a new feature called - Filtered Indexes
CREATE NONCLUSTERED INDEX fidx_abc_01
    ON t_abc ( EndDate )
WHERE EndDate > '20040401';
select * from t_abc where EndDate between '20050121' and '20050130';[url http://msdn.microsoft.com/en-us/library/cc280372.aspx]Filtered Indexes can be very helpful in both OLTP and DW. I'm wondering how to get the same thing done in Oracle.
Any suggestion beyond [url http://erturkdiriksoy.wordpress.com/2008/06/30/filtered-indexes-on-oracle/]Dunyada?

Answer 1

In 11.2, if you're building a data warehouse (and assuming that implies you have the partitioning option), you can declare that certain partitions of a local index are unusable which prevents Oracle from building them. If T_ABC were partitioned on END_DT, that would allow arbitrary queries where Oracle could perform partition pruning to make use of the index if it was available for those partitions without the need for an exact expression match.
Beyond that, if you are creating a function-based index, you would often want to create (or modify) the view that your application queries to specify the condition (i.e. a VW_RECENT_ABC that only has post-April, 2008 data) so that queries didn't have to specify the condition exactly.
Justin