Statistical maintenance functionality (AutoStats) in SQL Server

General information

SQL Server uses a cost-based optimizer that can be extremely sensitive to statistical information that is provided in tables and indexes. Without correct and up-to-date statistical information, SQL Server can find it difficult to determine the best execution plan for a particular query.

Statistics maintained on each table in SQL Server to aid the optimizer in cost-based decision making include the:

• Number of rows in the table.
• Number of pages used by the table.
• Number of modifications made to the keys of the table since the last update to the statistics.

Additional information is stored for indexes, including (for each index):

• An equi-height histogram on the first column.
• Densities on all column prefixes.
• Average key length.

Statistics on indexes are automatically created whenever a new index is built. In addition, it is possible to create and maintain statistics on other columns as well.

To maintain the statistical information as up-to-date as possible, SQL Server introduced AutoStat, which, through SQL Server's monitoring of table modifications, is capable of automatically updating the statistics for a table when a certain change threshold has been reached. Additionally, SQL Server introduced auto-create-statistics, which causes the server to automatically generate all statistics required for the accurate optimization of a specific query.

Determining when AutoStat generation is imminent

As mentioned above, AutoStat automatically updates the statistics for a particular table when a "change threshold" has been reached. The sysindexes.rowmodctr column maintains a running total of all modifications to a table that can adversely affect the query processor's decision-making process over time. This counter is updated each time any of the following events occurs:

• A single row insert is made.
• A single row delete is made.
• An update to an indexed column is made.

Note TRUNCATE TABLE does not update rowmodctr.

After table statistics have been updated, the rowmodctr value is reset to 0, and the table's statistics schema version is updated.

Further, in situations in which a stored procedure's execution plan is taken from cache and that plan is sensitive to statistics, the statistics schema version is compared to the current version. If there are new statistics available, the plan for the stored procedure will be recompiled.

The basic algorithm for auto update statistics is:

• If the cardinality for a table is less than 6 and the table is in the tempdb database, auto update occurs with every 6 modifications to the table.
• If the cardinality for a table is greater than 6 but less than or equal to 500, update occurs every 500 modifications.
• If the cardinality for a table is greater than 500, update statistics when (500 + 20 percent of the table) changes have occurred.
• For table variables, cardinality changes does not trigger auto update statistics.

Note In this strictest sense, SQL Server counts cardinality as the number of rows in the table.

Note In addition to cardinality, the selectivity of the predicate also affects Autostats generation. This means that statistics may not be updated after every 500 modifications if cardinality is less than 500, or for every 20 percent of changes if cardinality is greater than 500. A scale-up factor (the value ranges from 1 to 4) is generated, depending on the selectivity, and a product of this factor and the number of changes as obtained from the algorithm would be the actual number of modifications required for AutoStats generation.

The above algorithm can be summarized in the form of a table:

_________________________________________________________________________________
 Table Type | Empty Condition | Threshold When Empty |Threshold When Not Empty 
_________________________________________________________________________________
 Permanent  | < 500 rows      | # of Changes >= 500  | # of Changes >= 500 + (20% of Cardinality)
___________________________________________________________________________
 Temporary  | < 6 rows        | # of Changes >= 6    | # of Changes >= 500 + (20% of Cardinality)
___________________________________________________________________________
Table
Variables   | Change in cardinality does not affect AutoStats generation.
___________________________________________________________________________ 

The following two examples help demonstrate this concept.

Example 1

Consider the authors table in the pubs database, which contains 23 rows and has 2 indexes. The unique clustered index, UPKCL_auidind, is indexed on one column, au_id, and a composite nonclustered index, aunmind, has been created on the au_lname and au_fname columns. Because this table contains fewer than 500 rows, AutoStat will begin after 500 changes to the table data have occurred. The changes can be 500 or more inserts, deletes, changes to an indexed column such as au_lname, or any combination of these.

You can, therefore, predict when UPDATE STATISTICS will be initiated by monitoring the sysindexes.rowmodctr value, which will be incremented upon each update. When it reaches 500, you can expect UPDATE STATISTICS to start.

Example 2

Consider a second table, t2, that has a cardinality of 1,000. For tables with more than 500 rows, SQL Server will UPDATE STATISTICS when (500 + 20 percent) changes have been made. Doing the math, 20 percent of 1,000 is 200, so you can expect to see AutoStat start after approximately 700 modifications have been made to the table.

Automating AutoStats determination

To automate the determination of when AutoStat will run, you can poll the sysindexes table and identify when table modifications are reaching the starting point. The following is a basic algorithm for doing so:
 
You could later schedule a job to do the following:

• Run UPDATE STATISTICS against all tables that you were forced to disable during the day.
  
  -AND-
• Re-enable AutoStat, because each table's modification counter will have been reset to 0 when UPDATE STATISTICS was run.

Controlling whether UPDATE STATISTICS are run against a table

The most obvious solution to this question, when AutoStat has proven to be problematic, is to disable auto statistic generation, thereby leaving database administrators free to schedule UPDATE STATISTICS during less-intrusive times. You can do this by using the UPDATE STATISTICS statement or the sp_autostats stored procedure. The syntax for the UPDATE STATISTICS statement is:
The syntax for the sp_autostats stored procedure is: where <stats_flag> is "on" or "off".

You can also use sp_dboption to disable the automatic occurrence of UPDATE STATISTICS or CREATE STATISTICS on a per-database level:

Controlling the number of concurrent UPDATE STATISTICS processes

Currently, short of disabling AutoStat for specific tables, it is not possible to configure the number of automatic UPDATE STATISTICS statements that are being run concurrently. The server does, however, limit the number of concurrent UPDATE STATISTICS processes to four per processor.

Determining when Autostats are being run

You can use trace flag 205 to report when a statistics-dependent stored procedure is being recompiled as a result of AutoStat. This trace flag will write the following messages to the error log: When trace flag 205 is enabled, the following message will also bracket the AutoStat message from 8721 when statistics are updated. The opening message of the bracket can be distinguished by the RowModCnt value, which will be greater than 0. The closing bracket, after the UPDATE STATISTICS, will have a RowModCnt value of 0: For this message, "RowModCnt" is the total number of modifications to the table. "RowModLimit" is the threshold which, when exceeded, results in an UPDATE STATISTICS statement execution for the table.

It is also possible to enable trace flag 8721, which will dump information into the error log when AutoStat has been run. The following is an example of the type of message that you can expect to see: For this message, "Mods" is the total number of modifications to the table, "Bound" is the modification threshold, "Duration" is the amount of time that the UPDATE STATISTICS statement required to complete, and "UpdCount" is the count of updated statistics.

You can also use the SQL Server Profiler to identify when UPDATE STATISTICS statements are being run. To do this, perform the following steps:

1. On the Profiler menu, click Tools, and then click Options.
2. On the General tab, go to Events, and then select All Event Classes.
3. Define a new trace, and under Events, select Misc, select the Auto-Update Stats sub-event.

Note If many statistics are being updated by AutoStat, a great number of messages can be written to the error log. Thoroughly experiment with these trace flags before using them on any production or otherwise critical server.

Schema locks

SQL Server employs two types of schema locks, both of which are taken when it updates the statistics for a table:

   Sch-S: Schema Stability Lock
   ----------------------------
   This lock ensures that a schema element, such as a table or index, will
   not be dropped while any session holds a schema stability lock on the
   schema element.

   Sch-M-UPD-STATS: Schema Modification Lock
   -----------------------------------------
   This is a non-blocking lock that is used by the system to ensure that
   only one automatic UPDATE STATISTICS process is run against a table at
   any given point in time. The sp_lock stored procedure will report this
   lock has having a type = TAB, resouce = UPD-STATS and mode = SCH-M.
 

You can view these locks by running sp_lock or by selecting from the syslockinfo table.