Author: Arif Chaudhary
Introduction:
The performance of your completed worksheet has you wondering whether it is taking a little longer than it should. You can assess your workbook using a performance recording to determine what's slowing it down. This is a smart idea if you intend to distribute or publish the workbook.
As you interact with a workbook, Tableau's Performance Recording feature logs performance data about significant events.
Creating a Performance Recording in Tableau Desktop:
To start the recording, Go to: Help > Settings and Performance > Start Performance Recording.
To stop the recording, Users then look at the temporary workbook containing the recording session's outcomes.
Go to: Help > Settings and Performance > Stop Performance Recording.
The performance recording analysis can be saved as a packed worksheet (.twbx) and sent to Tableau support for technical support and more analysis.
Interpret and Analyze the Performance Recording:
Performance Summary and Detailed Views are the two main dashboards seen in a performance recording worksheet. A high-level summary of the time-consuming events is provided by the Performance Summary dashboard. When creating workbooks, advanced users should select the Detailed Views dashboard because it offers a lot more information.
1) Performance Summary
The dashboard for the performance summary contains three views: Timeline, Events, and Query.
1.1] Timeline
The performance summary recording dashboard's top view displays the events that took place while recording, organized chronologically from left to right. The seconds since Tableau started are shown on the bottom axis. The Workbook, Dashboard, and Worksheet dimensions in the timeline view provide the context of the event.
1.2] Events
This worksheet's aim is to provide a list of the activities that took place during the recording, arranged by duration, longest to shortest. This will assist you in prioritizing important problems that need to be solved right now.
Different colors are used to signify each event (which we can see on the right side of the above snippet).
1.2.1] Computing Layouts
The layouts processing is too slow. Simplify worksheet.
1.2.2] Compiling query
This event records the time that Tableau spent generating the queries. The complexity of the queries generated is shown by long compilation query times. Complexity could be because of using too many filters, complex worksheets, etc.
1.2.3] Executing Query
The underlying data format may not be optimized for Tableau, which is the most frequent reason why live connection queries take too long to process. Avoid live connections whenever possible, and employ extracts and similar performance-enhancing features.
1.2.4] Rendering
Running more VizQL Server processes on more machines will speed up server rendering.
1.2.5] Geocoding
The number of data points, or marks, is also a factor in the most frequent reason for bad performance. Think about utilizing less data or domain-specific data filtering. Filtering is one of the most frequent causes of bad performance, so keep it lean.
1.3] Query
The content for the query is shown in the Query section if you click on an Executing Query event in the Timeline or Events section of a performance summary dashboard.
2) Detailed Views
The dashboard for the detailed views contains four views: Depth, Exclusive CPU, Inclusive CPU, and Elapsed Time.
2.1] Depth
The Depth view, which is the top view in the Detailed Views dashboard, gives information on what occurs when a request is submitted. When limited to a single user request, this view is the most helpful.
2.2] Exclusive CPU, Inclusive CPU, and Elapsed Time:
For each activity, the Exclusive CPU, Inclusive CPU & Elapsed Time views offer aggregate statistics. The Count column displays the total number of times an activity occurred, and the bar chart displays the total amount of time spent on a single activity.
Conclusion:
In conclusion, the cornerstone for diagnosing Tableau performance issues is the performance recording & the associated dashboard. Keep your test case simple. Always examine the queries to gain a better understanding of what transpired. It will identify the connections between your logical and physical designs, the features in your dashboards and views, and the final SQL produced by Tableau.
References:
[1] https://help.tableau.com/current/pro/desktop/en-us/perf_record_create_desktop.htm