Tuning the Performance of the Tableau Server

When you want to remove the performance problems in the Tableau you always have to start the procedure by comparing the working performance of the same workbooks in the Tableau desktop. If the performance is good with the tableau desktop, then obviously the tableau server performance will be good too. If the performance on the desktop faces delay, then obviously there will be a delay in the Tableau server too.

So make sure that the Tableau desktop works well. Because once the Tableau desktop works well, the Tableau server can then be concentrated on. For better Tableau administration, the following factors need proper attention. Usually every server is different as a lot of factors influence its performance. However talking about Tableau Server, its performance can be balanced and improved by following basic ways:

• By configuring e-mail alerts in Tableau Server
• By collecting and analyzing Tableau Server data
• By adjusting the tasks and process configurations in Tableau Server
• By determining the performance bottlenecks

Watch this video on Tableau for Beginners:

Loading of Views in Tableau Server

Views are made to access Tableau via a browser. The client request comes to the Apache server and then goes to the Application server wgserver.exe. This server is responsible for browsing and Tableau interface. The request is then transferred to the VizQL process and then it goes to the data source. This happens once the view is on.

The views are dynamic tools which are made to for asking questions about the data. Hence, for all the communication of the views you need to also interact with the data.  Since it helps interact with the data, so the Tableau server can be considered as a data analytics and query tool. It is communicative, so the request sending delay will depend on the query delay.

Improving the performance of Tableau Server

Model Cache of the Queries

When the browser asks for the view, the server finds out if the pre-display calculations are already done. If it is already done, then the view is given to the browser directly. If those pre-display calculations are yet to be done, then the Tableau server does the calculations for display first and then the view is given to the browser.

The memory Cache of the Queries

If the view of browser has asked for was never asked by any other browser before, then the server will first check if there are any matching query of a prior time. If queries are there, then they are used to create the view and then the created view is sent to the browser. It is faster than the Model Cache but a bit costly.

Find out the total quantity of the Application Server Processes:

With a large number of Application server processes, the requests of views made by the browsers will hardly get connected to the request processor and data cache. So the number of applications needs to be reduced, which will allow the better Tableau server performance.

Balance the load by adding workers

During low load, the requests are faster and the speed declines as the load increases. Clustered nodes are created and added which are called as workers. Workers are added just to obtain a higher scalability, but workers will not help improve the performance of Tableau server. It will make the requests stay balanced. When there are a multiple numbers of servers in the environment, then interaction will require more processes between the servers. So when the individual server is not able to handle the load, then the workers are added to take the load.

We should add workers if and only if the below-mentioned conditions are met:

• If extra Application server processes cannot be added and if the engine is already filled with the traffic of the memory, input-output devices, APU.
• If the hardware cannot be again worked on for the betterment and if the machines in the server contain no other applications, then no workers can be added to the Tableau server environment.

Conclusion

After testing the server, the performance of Tableau server is likely to be enhanced. Please try the steps I have mentioned in this article and let us know whether it has helped you or not. If you guys have any other ideas, please do share by commenting in the boxes below.