Viewing and Analyzing Traces with SQL Server Profiler

  • Article

Use SQL Server Profiler to view captured event data in a trace. SQL Server Profiler displays data based on defined trace properties. One way to analyze SQL Server data is to copy the data to another program, such as SQL Server 2005 or Database Engine Tuning Advisor. Database Engine Tuning Advisor can use a trace file that contains SQL batch and remote procedure call (RPC) events if the Text data column is included in the trace. For more information, see Tuning the Physical Database Design. To make sure that the correct events and columns are captured for use with Database Engine Tuning Advisor, use the predefined Tuning template that is supplied with SQL Server Profiler.

When you open a trace by using SQL Server Profiler, the trace file does not need to have the .trc file extension if the file was created by either SQL Server Profiler or SQL Trace system stored procedures.

Note

SQL Server Profiler can also read SQL Trace .log files and generic SQL script files. When opening a SQL Trace .log file that does not have a .log file extension, such as trace.txt, specify SQLTrace_Log as the file format.

You can configure the SQL Server Profiler date and time display format to assist in trace analysis.

Troubleshooting Data

Using SQL Server Profiler, you can troubleshoot data by grouping traces or trace files by the Duration, CPU, Reads, or Writes data columns. Examples of data you might troubleshoot are queries that perform poorly or that have exceptionally high numbers of logical read operations.

Additional information can be found by saving traces to tables and using Transact-SQL to query the event data. For example, to determine which SQL:BatchCompleted events had excessive wait time, execute the following:

SELECT  TextData, Duration, CPU
FROM    trace_table_name
WHERE   EventClass = 12 -- SQL:BatchCompleted events
AND     CPU < (Duration * 1000)

Note

In SQL Server 2005, the server reports the duration of an event in microseconds (one millionth, or 10-6, of a second) and the amount of CPU time used by the event in milliseconds (one thousandth, or 10-3, of a second). In SQL Server 2000, the server reported both duration and CPU time in milliseconds. In SQL Server 2005, the SQL Server Profiler graphical user interface displays the Duration column in milliseconds by default, but when a trace is saved to either a file or a database table, the Duration column value is written in microseconds.

Displaying Object Names When Viewing Traces

If you wish to display the name of an object rather than the object identifier (Object ID), you must capture the Server Name and Database ID data columns along with the Object Name data column.

If you choose to group by the Object ID data column, make sure you group by the Server Name and Database ID data columns first, and then by the Object ID data column. Similarly, if you choose to group by the Index ID data column, make sure you group by the Server Name, Database ID, and Object ID data columns first, and then by the Index ID data columns. You must group in this order because object and index IDs are not unique among servers and databases (and among objects for index IDs).

Finding Specific Events Within a Trace

To find and group events in a trace, follow these steps:

  1. Create your trace.
    • When defining the trace, capture the Event Class, ClientProcessID, and Start Time data columns in addition to any other data columns you want to capture. For more information, see How to: Create a Trace (SQL Server Profiler).
    • Group the captured data by the Event Class data column, and capture the trace to a file or table. To group the captured data, click Organize Columns on the Events Selection tab of the Trace Properties dialog box. For more information, see How to: Organize Columns Displayed in a Trace (SQL Server Profiler).
    • Start the trace and stop it after the appropriate time has passed or number of events have been captured.
  2. Find the target events.
    • Open the trace file or table, and expand the node of the desired event class; for example, Deadlock Chain. For more information, see How to: Open a Trace File (SQL Server Profiler) or How to: Open a Trace Table (SQL Server Profiler).
    • Search through the trace data until you find the events for which you are looking (use the Find command on the Edit menu of SQL Server Profiler to help you find values in the trace). Note the values in the ClientProcessID and Start Time data columns of the events you trace.
  3. Display the events in context.
    • Display the trace properties, and group by the ClientProcessID data column rather than by the Event Class data column.
    • Expand the nodes of each client process ID you want to view. Search through the trace manually, or use Find until you find the previously noted Start Time values of the target events. The events are displayed in chronological order with the other events that belong to each selected client process ID. For example, the Deadlock and Deadlock Chain events, captured within the trace, appear immediately after the SQL:BatchStarting events within the expanded client process ID.

The same technique can be used to find any grouped events. Once you have found the events you seek, group them by ClientProcessID, Application Name, or another event class to view related activity in chronological order.

See Also

Tasks

How to: View a Saved Trace (Transact-SQL)
How to: View Filter Information (SQL Server Profiler)
How to: View Filter Information (Transact-SQL)
How to: Open a Trace File (SQL Server Profiler)
How to: Open a Trace File (SQL Server Profiler)
How to: Open a Trace Table (SQL Server Profiler)

Other Resources

fn_trace_getinfo (Transact-SQL)

Help and Information

Getting SQL Server 2005 Assistance

Change History

Release History

12 December 2006
Changed content:
  • Corrected the code example in the section, "Troubleshooting Data" where Duration column units are used.

5 December 2005
New content:
  • Added note at the end of the section, "Troubleshooting Data," which discusses the change from milliseconds to microseconds for the Duration column in trace events.
Changed content:
  • Corrected the code example in the section, "Troubleshooting Data," to compensate for the change in Duration column units (from milliseconds to microseconds).