3 Replies Latest reply on Nov 2, 2012 6:10 AM by Russell Christopher

    are data extracts processed on server?

    Ronak Patel

      I have some dashboards which all are using data extracts. When viewing/developing those dahsboards, the process is quite slow. When using data extracts, does the processing happen on the server or locally on the user side? I think the data extract is only 200mb too.

        • 1. Re: are data extracts processed on server?
          Russell Christopher

          Hi Ronak -


          Processing occurs on the Server. What is the difference in response for the same report when you are working with Desktop?

          • 2. Re: are data extracts processed on server?
            Ronak Patel

            It runs really slow  as well in the server, but I think I found out where the problem comes from. With the combination of certain filters, tableau is plotting about 1.2 million rows onto a heatmap and a table on the same dashboard. Is this something that will just cause an issue in tableau (plotting millions of data points) or can this be fixed by just increasing server specs?

            • 3. Re: are data extracts processed on server?
              Russell Christopher



              A few thoughts:


              • Whenever you have a slow report in Tableau Server, test it in Desktop - more often then not, it'll be slow there too...indicating you have an issue with a query which is slow to return data (most often) or that we're having to do lots and lots of local processing (less often - and YOUR scenario).
              • I never plot more than about 30-50K marks (rows) in a single report if I can help it. Why?
                • The human brain can't easily differentiate amongst so much "stuff" - If I present 100K marks to the user, the brain is going into "reading mode" vs. "seeing mode". We want the brain in "seeing  by data" mode as much as possible
                • Plotting more than a couple hundred thousand marks will tend to slow Tableau down. A million? Even more so.


              My advice would be to aggregate the data a little bit more if you can, or show an initial report that shows aggregated data a user can "explore". By clicking on something of interest in that "explore" report,  they'll then be presented with the most granular data in a "details" report - but much, much less of it because it'll already be filtered by something else.


              Hope this helps!