dta utility

Applies to: SQL Server

The dta utility is the command prompt version of Database Engine Tuning Advisor. The dta utility is designed to allow you to use Database Engine Tuning Advisor functionality in applications and scripts.

Note

The Database Engine Tuning Advisor is not supported for Azure SQL Database or Azure SQL Managed Instance. Instead, consider the strategies recommended in Monitoring and performance tuning in Azure SQL Database and Azure SQL Managed Instance. For Azure SQL Database, see also the Database Advisor performance recommendations for Azure SQL Database.

Like Database Engine Tuning Advisor, the dta utility analyzes a workload and recommends physical design structures to improve server performance for that workload. The workload can be a plan cache, a SQL Server Profiler trace file or table, or a Transact-SQL script. Physical design structures include indexes, indexed views, and partitioning. After analyzing a workload, the dta utility produces a recommendation for the physical design of databases and can generate the necessary script to implement the recommendation. Workloads can be specified from the command prompt with the -if or the -it argument. You can also specify an XML input file from the command prompt with the -ix argument. In that case, the workload is specified in the XML input file.

Syntax


dta  
[ -? ] |  
[  
      [ -S server_name[ \instance ] ]  
      { { -U login_id [-P password ] } | -E  }  
      { -D database_name [ ,...n ] }  
      [ -d database_name ]   
      [ -Tl table_list | -Tf table_list_file ]  
      { -if workload_file | -it workload_trace_table_name  |   
        -ip | -iq }  
      { -ssession_name | -IDsession_ID }  
      [ -F ]  
      [ -of output_script_file_name ]  
      [ -or output_xml_report_file_name ]  
      [ -ox output_XML_file_name ]  
      [ -rl analysis_report_list [ ,...n ] ]  
      [ -ix input_XML_file_name ]  
      [ -A time_for_tuning_in_minutes ]  
      [ -n number_of_events ]
      [ -l time_window_in_hours ]  
      [ -m minimum_improvement ]  
      [ -fa physical_design_structures_to_add ]  
      [ -fi filtered_indexes]  
      [ -fc columnstore_indexes]  
      [ -fp partitioning_strategy ]  
      [ -fk keep_existing_option ]  
      [ -fx drop_only_mode ]  
      [ -B storage_size ]  
      [ -c max_key_columns_in_index ]  
      [ -C max_columns_in_index ]  
      [ -e | -e tuning_log_name ]  
      [ -N online_option]  
      [ -q ]  
      [ -u ]  
      [ -x ]  
      [ -a ]  
]  

Arguments

-?
Displays usage information.

-A time_for_tuning_in_minutes
Specifies the tuning time limit in minutes. dta uses the specified amount of time to tune the workload and generate a script with the recommended physical design changes. By default dta assumes a tuning time of 8 hours. Specifying 0 allows unlimited tuning time. dta might finish tuning the entire workload before the time limit expires. However, to make sure that the entire workload is tuned, we recommend that you specify unlimited tuning time (-A 0).

-a
Tunes workload and applies the recommendation without prompting you.

-B storage_size
Specifies the maximum space in megabytes that can be consumed by the recommended index and partitioning. When multiple databases are tuned, recommendations for all databases are considered for the space calculation. By default, dta assumes the smaller of the following storage sizes:

  • Three times the current raw data size, which includes the total size of heaps and clustered indexes on tables in the database.

  • The free space on all attached disk drives plus the raw data size.

The default storage size does not include nonclustered indexes and indexed views.

-C max_columns_in_index
Specifies the maximum number of columns in indexes that dta proposes. The maximum value is 1024. By default, this argument is set to 16.

-c max_key_columns_in_index
Specifies the maximum number of key columns in indexes that dta proposes. The default value is 16, the maximum value allowed. dta also considers creating indexes with included columns. Indexes recommended with included columns may exceed the number of columns specified in this argument.

-D database_name
Specifies the name of each database that is to be tuned. The first database is the default database. You can specify multiple databases by separating the database names with commas, for example:

dta -D database_name1, database_name2...  

Alternatively, you can specify multiple databases by using the -D argument for each database name, for example:

dta -D database_name1 -D database_name2... n  

The -D argument is mandatory. If the -d argument has not been specified, dta initially connects to the database that is specified with the first USE database_name clause in the workload. If there is not explicit USE database_name clause in the workload, you must use the -d argument.

For example, if you have a workload that contains no explicit USE database_name clause, and you use the following dta command, a recommendation will not be generated:

dta -D db_name1, db_name2...  

But if you use the same workload, and use the following dta command that uses the -d argument, a recommendation will be generated:

dta -D db_name1, db_name2 -d db_name1  

-d database_name
Specifies the first database to which dta connects when tuning a workload. Only one database can be specified for this argument. For example:

dta -d AdventureWorks2022 ...  

If multiple database names are specified, then dta returns an error. The -d argument is optional.

If you are using an XML input file, you can specify the first database to which dta connects by using the DatabaseToConnect element that is located under the TuningOptions element. For more information, see Database Engine Tuning Advisor.

If you are tuning only one database, the -d argument provides functionality that is similar to the -d argument in the sqlcmd utility, but it does not execute the USE database_name statement. For more information, see sqlcmd Utility.

-E
Uses a trusted connection instead of requesting a password. Either the -E argument or the -U argument, which specifies a login ID, must be used.

-e tuning_log_name
Specifies the name of the table or file where dta records events that it could not tune. The table is created on the server where the tuning is performed.

If a table is used, specify its name in the format: [database_name].[owner_name].table_name. The following table shows the default values for each parameter:

Parameter Default value Details
database_name database_name specified with the -D option
owner_name dbo owner_name must be dbo. If any other value is specified, then dta execution fails and it returns an error.
table_name None

If a file is used, specify .xml as its extension. For example, TuningLog.xml.

Note

The dta utility does not delete the contents of user-specified tuning log tables if the session is deleted. When tuning very large workloads, we recommend that a table be specified for the tuning log. Since tuning large workloads can result in large tuning logs, the sessions can be deleted much faster when a table is used.

-F
Permits dta to overwrite an existing output file. If an output file with the same name already exists and -F is not specified, dtareturns an error. You can use -F with -of, -or, or -ox.

-fa physical_design_structures_to_add
Specifies what types of physical design structures dta should include in the recommendation. The following table lists and describes the values that can be specified for this argument. When no value is specified, dta uses the default -fa IDX.

Value Description
IDX_IV Indexes and indexed views.
IDX Indexes only.
IV Indexed views only.
NCL_IDX Nonclustered indexes only.

-fi
Specifies that filtered indexes be considered for new recommendations. For more information, see Create Filtered Indexes.

-fc
Specifies that columnstore indexes be considered for new recommendations. DTA will consider both clustered and nonclustered columnstore indexes. For more information, see
Columnstore index recommendations in Database Engine Tuning Advisor (DTA).

Applies to: SQL Server 2016 (13.x) and later.

-fk keep_existing_option
Specifies what existing physical design structures dta must retain when generating its recommendation. The following table lists and describes the values that can be specified for this argument:

Value Description
NONE No existing structures
ALL All existing structures
ALIGNED All partition-aligned structures.
CL_IDX All clustered indexes on tables
IDX All clustered and nonclustered indexes on tables

-fp partitioning_strategy
Specifies whether new physical design structures (indexes and indexed views) that dta proposes should be partitioned, and how they should be partitioned. The following table lists and describes the values that can be specified for this argument:

Value Description
NONE No partitioning
FULL Full partitioning (choose to enhance performance)
ALIGNED Aligned partitioning only (choose to enhance manageability)

ALIGNED means that in the recommendation generated by dta every proposed index is partitioned in exactly the same way as the underlying table for which the index is defined. Nonclustered indexes on an indexed view are aligned with the indexed view. Only one value can be specified for this argument. The default is -fp NONE.

-fx drop_only_mode
Specifies that dta only considers dropping existing physical design structures. No new physical design structures are considered. When this option is specified, dta evaluates the usefulness of existing physical design structures and recommends dropping seldom used structures. This argument takes no values. It cannot be used with the -fa, -fp, or -fk ALL arguments

-ID session_ID
Specifies a numerical identifier for the tuning session. If not specified, then dta generates an ID number. You can use this identifier to view information for existing tuning sessions. If you do not specify a value for -ID, then a session name must be specified with -s.

-ip
Specifies that the plan cache be used as the workload. The top 1,000 plan cache events for explicitly selected databases are analyzed. This value can be changed using the -n option.

-iq
Specifies that the Query Store be used as the workload. The top 1,000 events from the Query Store for explicitly selected databases are analyzed. This value can be changed using the -n option. See Query Store and Tuning Database Using Workload from Query Store for more information.

Applies to: SQL Server 2016 (13.x) and later.

-if workload_file
Specifies the path and name of the workload file to use as input for tuning. The file must be in one of these formats: .trc (SQL Server Profiler trace file), .sql (SQL file), or .log ( SQL Server trace file). Either one workload file or one workload table must be specified.

-it workload_trace_table_name
Specifies the name of a table containing the workload trace for tuning. The name is specified in the format: [database_name].[owner_name].table_name.

The following table shows the default values for each:

Parameter Default value
database_name database_name specified with -D option.
owner_name dbo.
table_name None.

Note

owner_name must be dbo. If any other value is specified, execution of dta fails and an error is returned. Also note that either one workload table or one workload file must be specified.

-ix input_XML_file_name
Specifies the name of the XML file containing dta input information. This must be a valid XML document conforming to DTASchema.xsd. Conflicting arguments specified from the command prompt for tuning options override the corresponding value in this XML file. The only exception is if a user-specified configuration is entered in the evaluate mode in the XML input file. For example, if a configuration is entered in the Configuration element of the XML input file and the EvaluateConfiguration element is also specified as one of the tuning options, the tuning options specified in the XML input file will override any tuning options entered from the command prompt.

-k maxtotalindexes
Maximum number of indexes in the recommendation.

-K maxtotalindexes
Maximum number of indexes per table.

-m minimum_improvement
Specifies the minimum percentage of improvement that the recommended configuration must satisfy.

-N online_option
Specifies whether physical design structures are created online. The following table lists and describes the values you can specify for this argument:

Value Description
OFF No recommended physical design structures can be created online.
ON All recommended physical design structures can be created online.
MIXED Database Engine Tuning Advisor attempts to recommend physical design structures that can be created online when possible.

If indexes are created online, ONLINE = ON is appended to its object definition.

-n number_of_events
Specifies the number of events in the workload that dta should tune. If this argument is specified and the workload is a trace file that contains duration information, then dta tunes events in decreasing order of duration. This argument is useful to compare two configurations of physical design structures. To compare two configurations, specify the same number of events to be tuned for both configurations and then specify an unlimited tuning time for both also as follows:

dta -n number_of_events -A 0  

In this case, it is important to specify an unlimited tuning time (-A 0). Otherwise, Database Engine Tuning Advisor assumes an 8 hour tuning time by default.

-l time_window_in_hours
Specifies the time window (in hours) when a query must have executed for it to be considered by DTA for tuning when using -iq option (Workload from Query Store).

dta -iq -l 48  

In this case, DTA will use Query Store as the source of workload and only consider queries that have executed with the past 48 hours.

Applies to: SQL Server 2016 (13.x) and later.

-of output_script_file_name
Specifies that dta writes the recommendation as a Transact-SQL script to the file name and destination specified.

You can use -F with this option. Make sure that the file name is unique, especially if you are also using -or and -ox.

-or output_xml_report_file_name
Specifies that dta writes the recommendation to an output report in XML. If a file name is supplied, then the recommendations are written to that destination. Otherwise, dta uses the session name to generate the file name and writes it to the current directory.

You can use -F with this option. Make sure that the file name is unique, especially if you are also using -of and -ox.

-ox output_XML_file_name
Specifies that dta writes the recommendation as an XML file to the file name and destination supplied. Ensure that Database Engine Tuning Advisor has permissions to write to the destination directory.

You can use -F with this option. Make sure that the file name is unique, especially if you are also using -of and -or.

-P password
Specifies the password for the login ID. If this option is not used, dta prompts for a password.

-q
Sets quiet mode. No information is written to the console, including progress and header information.

-rl analysis_report_list
Specifies the list of analysis reports to generate. The following table lists the values that can be specified for this argument:

Value Report
ALL All analysis reports
STMT_COST Statement cost report
EVT_FREQ Event frequency report
STMT_DET Statement detail report
CUR_STMT_IDX Statement-index relations report (current configuration)
REC_STMT_IDX Statement-index relations report (recommended configuration)
STMT_COSTRANGE Statement cost range report
CUR_IDX_USAGE Index usage report (current configuration)
REC_IDX_USAGE Index usage report (recommended configuration)
CUR_IDX_DET Index detail report (current configuration)
REC_IDX_DET Index detail report (recommended configuration)
VIW_TAB View-table relations report
WKLD_ANL Workload analysis report
DB_ACCESS Database access report
TAB_ACCESS Table access report
COL_ACCESS Column access report

Specify multiple reports by separating the values with commas, for example:

... -rl EVT_FREQ, VIW_TAB, WKLD_ANL ...  

-S server_name[ \instance]
Specifies the name of the computer and instance of SQL Server to connect to. If no server_name is specified, dta connects to the default instance of SQL Server on the local computer. This option is required when connecting to a named instance or when executing dta from a remote computer on the network.

-s session_name
Specifies the name of the tuning session. This is required if -ID is not specified.

-Tf table_list_file
Specifies the name of a file containing a list of tables to be tuned. Each table listed within the file should begin on a new line. Table names should be qualified with three-part naming, for example, AdventureWorks2022.HumanResources.Department. Optionally, to invoke the table-scaling feature, the name of an existing table can be followed by a number indicating the projected number of rows in the table. Database Engine Tuning Advisor takes into consideration the projected number of rows while tuning or evaluating statements in the workload that reference these tables. Note that there can be one or more spaces between the number_of_rows count and the table_name.

This is the file format for table_list_file:

database_name.[schema_name].table_name [number_of_rows]

database_name.[schema_name].table_name [number_of_rows]

database_name.[schema_name].table_name [number_of_rows]

This argument is an alternative to entering a table list at the command prompt (-Tl). Do not use a table list file (-Tf) if you are using -Tl. If both arguments are used, dta fails and returns an error.

If the -Tf and -Tl arguments are omitted, all user tables in the specified databases are considered for tuning.

-Tl table_list
Specifies at the command prompt a list of tables to be tuned. Place commas between table names to separate them. If only one database is specified with the -D argument, then table names do not need to be qualified with a database name. Otherwise, the fully qualified name in the format: database_name.schema_name.table_name is required for each table.

This argument is an alternative to using a table list file (-Tf). If both -Tl and -Tf are used, dta fails and returns an error.

-U login_id
Specifies the login ID used to connect to SQL Server.

-u
Launches the Database Engine Tuning Advisor GUI. All parameters are treated as the initial settings for the user interface.

-x
Starts tuning session and exits.

Remarks

Press CTRL+C once to stop the tuning session and generate recommendations based on the analysis dta has completed up to this point. You will be prompted to decide whether you want to generate recommendations or not. Press CTRL+C again to stop the tuning session without generating recommendations.

Examples

A. Tune a workload that includes indexes and indexed views in its recommendation

This example uses a secure connection (-E) to connect to the tpcd1G database on MyServer to analyze a workload and create recommendations. It writes the output to a script file named script.sql. If script.sql already exists, then dta will overwrite the file because the -F argument has been specified. The tuning session runs for an unlimited length of time to ensure a complete analysis of the workload (-A 0). The recommendation must provide a minimum improvement of 5% (-m 5). dta should include indexes and indexed views in its final recommendation (-fa IDX_IV).

dta -S MyServer -E -D tpcd1G -if tpcd_22.sql -F -of script.sql -A 0 -m 5 -fa IDX_IV  

B. Limit disk use

This example limits the total database size, which includes the raw data and the additional indexes, to 3 gigabytes (GB) (-B 3000) and directs the output to d:\result_dir\script1.sql. It runs for no more than 1 hour (-A 60).

dta -D tpcd1G -if tpcd_22.sql -B 3000 -of "d:\result_dir\script1.sql" -A 60  

C. Limit the number of tuned queries

This example limits the number of queries read from the file orders_wkld.sql to a maximum of 10 (-n 10) and runs for 15 minutes (-A 15), whichever comes first. To make sure that all 10 queries are tuned, specify an unlimited tuning time with -A 0. If time is important, specify an appropriate time limit by specifying the number of minutes that are available for tuning with the -A argument as shown in this example.

dta -D orders -if orders_wkld.sql -of script.sql -A 15 -n 10  

D. Tune specific tables listed in a file

This example demonstrates the use of table_list_file (the -Tf argument). The contents of the file table_list.txt are as follows:

AdventureWorks2022.Sales.Customer  100000  
AdventureWorks2022.Sales.Store  
AdventureWorks2022.Production.Product  2000000  

The contents of table_list.txt specifies that:

  • Only the Customer, Store, and Product tables in the database should be tuned.

  • The number of rows in the Customer and Product tables are assumed to be 100,000 and 2,000,000, respectively.

  • The number of rows in Store are assumed to be the current number of rows in the table.

    Note that there can be one or more spaces between the number of rows count and the preceding table name in the table_list_file.

    The tuning time is 2 hours (-A 120) and the output is written to an XML file (-ox XMLTune.xml).

dta -D pubs -if pubs_wkld.sql -ox XMLTune.xml -A 120 -Tf table_list.txt  

See also