85 lines
5.4 KiB
HTML
85 lines
5.4 KiB
HTML
<?xml version="1.0" encoding="UTF-8"?>
|
|
<!DOCTYPE html
|
|
PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">
|
|
<html lang="en-us" xml:lang="en-us">
|
|
<head>
|
|
<meta http-equiv="Content-Type" content="text/html; charset=utf-8" />
|
|
<meta name="security" content="public" />
|
|
<meta name="Robots" content="index,follow" />
|
|
<meta http-equiv="PICS-Label" content='(PICS-1.1 "http://www.icra.org/ratingsv02.html" l gen true r (cz 1 lz 1 nz 1 oz 1 vz 1) "http://www.rsac.org/ratingsv01.html" l gen true r (n 0 s 0 v 0 l 0) "http://www.classify.org/safesurf/" l gen true r (SS~~000 1))' />
|
|
<meta name="DC.Type" content="reference" />
|
|
<meta name="DC.Title" content="Indexes versus column statistics" />
|
|
<meta name="abstract" content="If you are trying to decide whether to use statistics or indexes to provide information to the Statistics Manager, keep the following differences in mind." />
|
|
<meta name="description" content="If you are trying to decide whether to use statistics or indexes to provide information to the Statistics Manager, keep the following differences in mind." />
|
|
<meta name="DC.subject" content="statistics manager, statistics versus indexes, index" />
|
|
<meta name="keywords" content="statistics manager, statistics versus indexes, index" />
|
|
<meta name="DC.Relation" scheme="URI" content="statsmanager.htm" />
|
|
<meta name="copyright" content="(C) Copyright IBM Corporation 1998, 2006" />
|
|
<meta name="DC.Rights.Owner" content="(C) Copyright IBM Corporation 1998, 2006" />
|
|
<meta name="DC.Format" content="XHTML" />
|
|
<meta name="DC.Identifier" content="statsvindex" />
|
|
<meta name="DC.Language" content="en-us" />
|
|
<!-- All rights reserved. Licensed Materials Property of IBM -->
|
|
<!-- US Government Users Restricted Rights -->
|
|
<!-- Use, duplication or disclosure restricted by -->
|
|
<!-- GSA ADP Schedule Contract with IBM Corp. -->
|
|
<link rel="stylesheet" type="text/css" href="./ibmdita.css" />
|
|
<link rel="stylesheet" type="text/css" href="./ic.css" />
|
|
<title>Indexes versus column statistics</title>
|
|
</head>
|
|
<body id="statsvindex"><a name="statsvindex"><!-- --></a>
|
|
<!-- Java sync-link --><script language="Javascript" src="../rzahg/synch.js" type="text/javascript"></script>
|
|
<h1 class="topictitle1">Indexes versus column statistics</h1>
|
|
<div><p>If you are trying to decide whether to use statistics or indexes
|
|
to provide information to the Statistics Manager, keep the following differences
|
|
in mind.</p>
|
|
<div class="section"><p>One major difference between indexes and column statistics is
|
|
that indexes are permanent objects that are updated when changes to the underlying
|
|
table occur, while column statistics are not. If your data is constantly changing,
|
|
the Statistics Manager may need to rely on stale column statistics. However,
|
|
maintaining an index after each change to the table might take up more system
|
|
resources than refreshing the stale column statistics after a group of changes
|
|
to the table have occurred. </p>
|
|
</div>
|
|
<div class="section"><p>Another difference is the effect that the existence of new indexes
|
|
or column statistics has on the Optimizer. When new indexes become available,
|
|
the Optimizer will consider them for implementation. If they are candidates,
|
|
the Optimizer will re-optimize the query and try to find a better implementation.
|
|
However, this is not true for column statistics. When new or refreshed column
|
|
statistics are available, the Statistics Manager will interrogate immediately.
|
|
Reoptimization will occur only if the answers are significantly different
|
|
from the ones that were given before these refreshed statistics. This means
|
|
that it is possible to use statistics that are refreshed without causing a
|
|
reoptimization of an access plan.</p>
|
|
</div>
|
|
<div class="section"><p>When trying to determine the selectivity of predicates, the Statistics
|
|
Manager considers column statistics and indexes as resources for its answers
|
|
in the following order:</p>
|
|
</div>
|
|
<div class="section"> <ol><li>Try to use a multi-column keyed index when ANDed or ORed predicates reference
|
|
multiple columns</li>
|
|
<li>If there is no perfect index that contains all of the columns in the predicates,
|
|
it will try to find a combination of indexes that can be used.</li>
|
|
<li>For single column questions, it will use available column statistics</li>
|
|
<li>If the answer derived from the column statistics shows a selectivity of
|
|
less than 2%, indexes are used to verify this answer</li>
|
|
</ol>
|
|
</div>
|
|
<div class="section"><p>Accessing column statistics to answer questions is faster than
|
|
trying to obtain these answers from indexes.</p>
|
|
</div>
|
|
<div class="section"><p>Column statistics can only be used by SQE. For CQE, all statistics
|
|
are retrieved from indexes.</p>
|
|
</div>
|
|
<div class="section"><p>Finally, column statistics can be used only for query optimization.
|
|
They cannot be used for the actual implementation of a query, whereas indexes
|
|
can be used for both.</p>
|
|
</div>
|
|
</div>
|
|
<div>
|
|
<div class="familylinks">
|
|
<div class="parentlink"><strong>Parent topic:</strong> <a href="statsmanager.htm" title="As stated earlier, the collection of statistics is handled by a separate component called the Statistics Manager. Statistical information can be used by the query optimizer to determine the best access plan for a query. Since the query optimizer bases its choice of access plan on the statistical information found in the table, it is important that this information be current.">Collecting statistics with the Statistics Manager</a></div>
|
|
</div>
|
|
</div>
|
|
</body>
|
|
</html> |