2 Replies Latest reply on Jun 18, 2013 9:35 AM by Toby Erkson

    Tableau Data Engine Storage Best Practices


      Looking at Tableau Server Architecture I got that Data Engine is a component that stores Data Extracts in some proprietary storage. Some of my extracts can potentially take up to 50-100 GB. So I am trying to understand what is best environment that accommodate such extracts - i.e:

      1. Are extracts partially loaded to the RAM and so I need more RAM thrown to the server?
      2. Should I use High Performance Storage (i.e. SSD discs) or I will not benefit from that?


      Is there any information available about architecture of Tableau Data Engine Storage that can answer these questions, or probably someone from Tableau can help with that.

        • 1. Re: Tableau Data Engine Storage Best Practices
          Toby Erkson

          There are just a couple people here that are really knowledgable about Server so I hope they'll chime in.  I do know that the more RAM you can give your server the better!  I seriously doubt that only part of an extract would be loaded into RAM but I won't swear by it.


          What is your operating system?  How many cores?  How much RAM?  How much available hard drive space?

          Also, why are your extracts so large?  The extract should only have the needed data for the report and nothing more.  Hide fields not being used, filter the data down (for example, a WHERE clause if using Custom SQL), etc.

          • 2. Re: Tableau Data Engine Storage Best Practices
            Toby Erkson

            Okay, let's get your thread to a location so that others can see it.  Click on the "Move" link on the right side of your screen:


            Next, click "Spaces" and then "Forums":


            Click the orange button that shows up in the next pop-up and your thread will get moved to more eye-*****


            Don't worry, this wasn't directly your fault, the forum is just a little "special"

            P.S. Thanks to Shawn Wallwork for pointing out this thread.