Striving for Optimal Performance – Invisible Indexes and Hints

Invisible Indexes and Hints

In this post I would like to remove some misinformation about the utilization of hints with invisible indexes.

Let’s start by providing you what two sources say about that topic:

Oracle Database 11g: New Features for DBAs and Developers (by Sam R. Alapati and Charles Kim, page 132)
“If you want the optimizer to take the invisible index into account, you must use the index hint, as shown in the following example:“

Oracle Database 11g: The Top New Features for DBAs and Developers – Schema Management (by Arup Nanda)
“To make the optimizer use the index again, you have to explicitly name the index in a hint:”

Both sources point out that a hint can be used to compel the query optimizer to use an invisible index. Both also provide an example (not shown here). But, what is more important, both are wrong! In other words, a hint cannot be used to compel the query optimizer to use an invisible index.

What’s going on?!?

The problem is that this specific information used to be correct for the beta release of Oracle Database 11g. But that behavior was a bug, not a feature! Therefore, once the bug was fixed, this piece of information was no longer valid.

This is what happens when authors use beta releases for doing their tests…

SQL> EXPLAIN PLAN FOR SELECT * FROM t1 WHERE n = 1; SQL> SELECT * FROM table(dbms_xplan.display(format=>'basic')); PLAN_TABLE_OUTPUT ---------------------------------- Plan hash value: 3617692013 ---------------------------------- | Id | Operation | Name | ---------------------------------- | 0 | SELECT STATEMENT | | | 1 | TABLE ACCESS FULL| T1 | ---------------------------------- SQL> EXPLAIN PLAN FOR SELECT /*+ index(t1) */ * FROM t1 WHERE n = 1; SQL> SELECT * FROM table(dbms_xplan.display(format=>'basic')); PLAN_TABLE_OUTPUT ---------------------------------- Plan hash value: 3617692013 ---------------------------------- | Id | Operation | Name | ---------------------------------- | 0 | SELECT STATEMENT | | | 1 | TABLE ACCESS FULL| T1 | ---------------------------------- SQL> EXPLAIN PLAN FOR SELECT /*+ use_invisible_indexes */ * FROM t1 WHERE n = 1; SQL> SELECT * FROM table(dbms_xplan.display(format=>'basic')); PLAN_TABLE_OUTPUT -------------------------------------------- Plan hash value: 100931685 -------------------------------------------- | Id | Operation | Name | -------------------------------------------- | 0 | SELECT STATEMENT | | | 1 | TABLE ACCESS BY INDEX ROWID| T1 | | 2 | INDEX RANGE SCAN | I_N | --------------------------------------------

5 Comments

Radoslav Golian

November 13, 2008 at 15:15 | Permalink

Good remark, thank you. I didn’t know that this was a bug.

Arup Nanda

September 1, 2009 at 06:42 | Permalink

You are right, Chris. And I apologize for not responding to it earlier.
I was referred to this literally hours ago. Sometimes things change albeit
in small amounts between beta and production. Those articles were reviewed
by Oracle PM and they were supposed to have caught that. Well…. Anyway,
that’s not an excuse; but just a background.

In 11gR2, the series will be written with production code alone.

Christian Antognini

September 1, 2009 at 09:20 | Permalink

Hi Arup

I sympathize with that… I know that editors want to have books (in your case, articles) as soon as the production release is available and, therefore, authors are pushed to work with beta releases. Hence, there are good chances that something like that happens.

I originally wrote this post because I heard that specific “myth” too many times! I though it was good to stop it. Or, at least, to try…

It’s good to know that for 11gR2 it will be different.

Cheers,
Chris

Mark Lancaster

February 22, 2010 at 05:55 | Permalink

Hi Chris

An update on invisible indexes and hints.
Oracle introduced the USE_INVISIBLE_INDEXES hint in 11.1.0.6, so provided your patched up it works correctly.

e.g. select /*+ USE_INVISIBLE_INDEXES */ * from your_table;

In fact they added quite a few new hints, see the V$SQL_HINT view for details.

Regards

Mark

Christian Antognini

February 28, 2010 at 16:32 | Permalink

Hi Mark

Thank you very much for your comment. It points out an important thing…

To make things clear, in my post I was speaking about hints like INDEX and not the new hint USE_INVISIBLE_INDEXES. Let me give an example to make the point cristal clear to everyone

Cheers,
Chris

Invisible Indexes and Hints

5 Comments

Leave a Reply Cancel reply