Index Statistics in SQL Server 7.0

• Introduction
• Distribution Statistics
• View Distribution Statistics
• Update Distribution Statistics

Introduction

Sometimes it is difficult to determine which indexes to use when

processing a query. In this case, the query optimizer uses distribution

statistics.

SQL Server 7.0 can update distribution statistics automatically, so

in most cases it is not necessary to update distribution statistics

manually and you can rely on SQL Server decisions to update

distribution statistics.

In this article, I want to tell you about distribution statistics,

and about how you can view and update distribution statistics.

Distribution Statistics

In comparison with SQL Server 6.5, there are no distribution pages in

SQL Server 7.0: distribution statistics are stored as a long string of

bits across multiple pages in the same way image data is stored.

The pointer to to this distribution data is stored in the statblob

column in the sysindexes table.

The distribution statistics are used to determine which indexes to

use when processing a query or it is not possible to use an index and

the full table scan will be made.

SQL Server 7.0 can update distribution statistics automatically, so

it is not longer necessary to run UPDATE STATISTICS statement to

update statistics manually. The SQL Server decides to update

distribution statistics automatically by analyzing the value in the

rowmodctr column in the sysindexes table.

Now distribution statistics can store the statistics information not

only for the first column in index and not only for the indexed columns.

You can create the distribution statistics for unindexed columns with

the CREATE STATISTICS statement.

You can view or change the automatic UPDATE STATISTICS setting with

sp_autostats stored procedure. By default, SQL Server 7.0 will

automatically update distribution statistics.

View Distribution Statistics

You can use DBCC SHOW_STATISTICS statement to display the current

distribution statistics for the specified target on the specified

table.

This is the syntax:

DBCC SHOW_STATISTICS (table, target)

where

table

is the table with which the index is associated.

target

is the name of the object (index name or collection) for which to display statistics information.

This is the example to view the index statistics for the aunmind

index from the authors table in the pubs database:

USE pubs

GO
DBCC SHOW_STATISTICS (authors, aunmind)
GO

Update Distribution Statistics

To update distribution statistics, you can use UPDATE STATISTICS statement.

This is the syntax (from SQL Server Books Online):

UPDATE STATISTICS table

    [
        index
        | (statistics_name[,...n])
    ]
    [    WITH
        [
            [FULLSCAN]
            | SAMPLE number {PERCENT | ROWS}]
        ]
        [[,] [ALL | COLUMNS | INDEX]
        [[,] NORECOMPUTE]
    ]

This is the example to update distribution statistics for the

aunmind index from the authors table in the pubs database:

USE pubs

GO
UPDATE STATISTICS authors aunmind
GO

For more information, see: UPDATE STATISTICS (T-SQL)

Notes

1. If you set to not automatically recompute distribution statistics,

   then you should manually run UPDATE STATISTICS statement periodically

   (when a large amount of data in an indexed column has been added,

   changed, or deleted).

2. After running TRUNCATE TABLE statement, the distribution statistics

   will also be deleted, so if you set to not automatically recompute

   statistics, then after adding new data, you should manually run

   UPDATE STATISTICS statement to recreate distribution statistics.