1 2 Previous Next 24 Replies Latest reply on May 4, 2017 11:38 PM by Yuriy Fal

    Tableau Server Optimization in Extract Intensive Deployment

    Kiran Anumalla

      Hello everyone,

       

      This is regarding the Tableau Server configuration changes to improve the performance of the server –

       

      Problem Area:

      1. Slow performance in rendering the report while navigating/ drill-down/selecting filters

       

      Some Analysis about the System and server –

      System Configuration for all the nodes

      • 64 bit
      • 64 GB RAM
      • 16 cores processor

      Tableau Server version – 10.2.1

      Tableau Server components settings for 1 Primary + 2 Worker nodes -

       

      Current Tableau Server Load and usage–

      1. Around 150 users using the report from 10 to 6 weekdays
      2. Most of the reports are extract based. There are like billions of data in the extract. So, it is Extract Intensive Server
      3. Extracts are mostly refreshed during non-peak hours

       

      Some Tableau server components tasks -

      1. Data Engine: process stored extracts and answer queries (Need to focus on this component)
      2. VizQL server process handles loading and rendering views for Tableau Server users (Need to focus on this component)
      3. Background process refreshes extracts

       

      As per Tableau Server Administrator and Tableau Scalability guide, for Extract Intensive deployment, it is best to isolate data engine on its own node.

       

      Need Suggestions:

      1. What should be the proper server process configurations for the Primary and each of the worker nodes
      2. Also, we are open to add more worker nodes if needed and what should be the configuration of the overall the Primary and the worker nodes (Before this I need to be sure of Point 1 above)

       

      Note:

      Below are some of the performance tips that have already been incorporated -

      1. Database tuning tips like indexes, joins
      2. Extracting only required columns and rows
      3. Using Context filters, Assuming Referential Integrity and many more performance tuning tips while creating dashboards
      4. Refreshing extracts during off period

       

       

       

      Many thanks in advance

        1 2 Previous Next