Last Modified Date: 06 Sep 2017
- Tableau Server 10.0, 10.1, 10.2
Increase the number of Data Engine Processes on computers running Tableau ServerFor each computer in the Tableau Server cluster with Data Engine process, make sure there are at least 2 Data Engine processes. Any computer that has only one Data Engine process should be modified to include at least 2 Data Engine processes.
Additional Data Engine Optimizations
- For 10.0.0 - 10.0.4 and 10.1.0 - 10.1.2, increase the Data Engine processes to 3 or 4. Note: CPU load should be monitored closely as part of the configuration change.
- For 10.0.5 - 10.0.7 and 10.1.3 - 10.1.5, increase the Data Engine processes to 3 or 4 and run the steps below for optimal performance.
Optimization of data engine string:
- On the computer running Tableau Server, open the Command Prompt window as an administrator and run the following command to change the directory to the Tableau Server bin folder:
cd "C:\Program Files\Tableau\Tableau Server\[version]\bin"
- At the command prompt, run the following commands in order:
tabadmin set dataengine.optimize_icu true
For 10.0.9, 10.1.6 and newer versions, the data engine string optimization occurs by default and these additional steps are not necessary.
Other Optimizations for upgrades to major versions
As with any major version upgrade, we recommend optimizing existing workbooks and extracts for the new functionality.
- For more information about optimizing extracts, see Optimize Extracts.
- For more information about optimizing workbooks, see Optimize Workbook Performance.
CauseIn Tableau Server 10, Tableau reduced overall latency and throughput for views. This optimization can cause increased use of Data Engine. The increased volume of requests may result in some performance delays if the configuration is unchanged.
Additional InformationChanges to Data Engine will increase CPU consumption during peak extract use. This is expected result of the optimizations in Tableau Server 10. Review and test these changes to understand the impact to performance and CPU load before applying these changes to production.
Thank you for providing your feedback on the effectiveness of the article.
Open new Case
Training and Tutorials