When we talk about performance, we mean the speed the user experiences. That might mean performance of data analysis, or loading time of views or dashboards that we access on our Desktop or Server. Be aware that the Server is used for scalability and not performance; if it is slow on the Desktop, it will be slow on the Server.
If we work with hundreds of millions of records, designing an efficient workbook is critically important to the speed at which we can work with.So how can we make run our dashboards faster?
Tableau provides a number of features designed to help us optimize performance. The first step is to identify the troublesome points. Are the databases, data sources, data size, data connection, extracts, calculations, layout, filters/parameters the problem? By running a performance recording, we are able to identify the performance degradation causes and make the dashboard more performant. The performance recorder outputs a Tableau report showing you the query times for all your different elements. It pinpoints slow worksheets, slow queries, and long render-times on the dashboard. Therefore, running and interpreting a performance recording is a good and recommended start for our workbook/dashboard performance diagnoses.

Figure 1 Performance Recording Option (Help > Settings and Performance > Show Performance Recording) and an example result report.

Now that we know exactly which views or data connections are slowing the performance down, here is an additional checklistwith best practices to take into account when building our dashboard along with somequick solutions.
Performance-saving tricks and tips:
  • Data:
    • Use extracts vs. live connection
    • Minimize the number of fields → Use “Hide All Unused Fields” option.
    • Minimize the number of records → Use extract filters to keep only the data you need.

Figure 2 Hide All Unused Fields option (Data tab > “⌵” dropdown menu> Hide All Unused Fields)

  • Computations
    • Optimize our calculations → when possible:
      • Perform row-level calculations in the database
      • Avoid nested calculations
      • Reduce the granularity of LOD or table calculations
      • Use MIN or MAX instead of AVG
      • Make groups with calculations
      • Use Booleans or numeric calculations instead of string calculations
      • Use COUNT () instead of COUNTD() function
    • Limit your filters by number and type → Keep number of filters and parameters down.
      Note that inefficient and excessive filters are one of the most common causes of poorly performing workbooks and dashboards. Therefore, as powerful and expressive filtering in Tableau is, keeping the number of quick filters and parameters down is crucial for the performance of the Dashboard. So, what can we do to effectively restrict the amount of data for our analysis? Here is a number of easily applicable tricks to filter our data wisely and reduce the number of queries in our Dashboard:
  • Create data source filters (when possible). This will reduce the number of data in the data source
  • Use “Keep Only” filters instead of “Exclude” filters. Exclude filters run much slower than Keep Only filters as they load the entire domain of a dimension, while Keep Only filters do not.
  • Use context filters cautiously, for example to include only data of interest, and then set a numerical or top a N filter.
  • Use a continuous date filter instead of discrete date filters, they are faster.
  • Use numeric or Boolean filters. Computers process integers and true/false statements faster than strings.
  • Use parameters and action filters. These reduce the query load.
  • Use the APPLY button on a multi-select drop down filter. These reduce the query load as it will just run the query once the Apply button is clicked and not after every single selection.

Figure 3 Tableau filter order and Show Apply Button option (Filter menu> Customize > Show Apply Button)

  • Reduce the number of marks (data points) in the view → Reduce unneeded dimensions, try displaying your data in different types of views or compile related views and connect them with action filters.

Figure 4 Number of marks in a Worksheet

  • Layout:
    • Max. 4 worksheets per Dashboard
    • Dashboard size → Fixed. Selecting the automatic or range option on the size of the Dashboard means that Tableau will automatically scale the dashboard to fit each viewer´s screen or window. This will not only not have an unpredictable result when automatically scales appearance, but will also decrease the performance.
    • Reduce number of containers used in the Dashboard → clean up all the unnecessary containers and elements in the dashboard. To view the list of all items that are on the dashboard go to the Layout tab> Objects > “⌵” icon to unfold list.

Figure 5 Best practice Layout examples: Max number of sheets, fixed dashboard size and view of Dashboard items list

  • End up by Cleaning up the workbooks!
    • Delete or consolidate unused worksheets and data sources
      • Worksheets: Right click on dashboard tab > Hide All Sheet. Repeat for all dashboards in the workbook and delete the remaining worksheets.
      • Data sources: right click on data source > Close

 

For more details on these areas and many more, contact us on info@m2dot.com.
Stayed tuned, soon coming… how to create a Spiral with Polygon charts and introducing the new topic of AWS!
Cheers,
Your M2. Team