2 Replies Latest reply on May 18, 2016 7:41 AM by Steve Burger

    Tableau Online Data Fidelity Problems?

    Colin Jemmott

      I am the administrator for our Tableau Online account.  I am seeing an intermittent error that I think may be a bug on the Tableau Online side.  In short, sometimes when my coworker and I look at the same dashboard at the same time, we see different results.

       

      Specifically, earlier today I looked at a Dashboard where all of the data ended at 4pm.  At the same time my coworker logged on and some of the plots ended at 4pm and some cut off a few hours earlier on the same dashboard.  In one case the time slider for a line plot quick filter ended earlier than a table with the exact same data in it.

       

      Additional notes:

       

      - Same worksheet, data set, and we both have administrator privileges. 

      - This issue has been persistent, and I have seen it on other workbooks also. 

      - I am uploading updates via Tableau Desktop once per hour.  The file is large and sometimes I get errors, but retry.

      - Pressing the reload button in the workbook, the browser refresh, and even quitting and reopening the browser do not always solve this problem, though sometimes does.

      - Sometimes the time sliders, plots and table all show different end points.

       

      I am taking a shot in the dark here, but it feels like data is being cached somewhere and not refreshing properly.  Has anyone else seen this problem?  Is it possible I am doing something wrong?

        • 1. Re: Tableau Online Data Fidelity Problems?
          Toby Erkson

          No idea but I would suspect caching, on the Online Server and/or browser.  If you don't get an answer I would recommend working with Tableau Support.

          • 2. Re: Tableau Online Data Fidelity Problems?
            Steve Burger

            Colin,

             

            I second Toby's comment that it could be a matter of timing and caching from the Tableau Online instance. Especially given that you are running hourly refreshes.

             

            I don't think you are doing anything wrong if you see success more frequently than you experience these issues.


            Remember that the data refresh happens within a window and not necessarily exactly at the same time, for the same duration, every hour.

             

            If it's a persistent issue, yes please create a support ticket either through your customer portal or support.tableau.com. If for no other reason, so our engineers can be made aware and the issue documented.

             

            Regarding your comment: "I am uploading updates via Tableau Desktop once per hour.  The file is large and sometimes I get errors, but retry." Are you using a full extract refresh or an incremental?


            If you are running a full extract refresh hourly, and starting to experience time-out issues, or the refresh not completing successful and you suspect the file size is the issue, time to think about pro's and con's of making the hourly refresh incremental, saving a full refresh for something like a weekend overnight.

             

            In addition, there are other strategies to reduce the size of your extract (and subsequent refresh) such as hiding unused dimensions and aggregating the extract data only to the visible dimensions in your views. (Decisions you make at the time of creating a new extract).

             

            Best resource for more info on that topic is here:

            Quick Start: Incremental Refresh

            Quick Start: Incremental Extract Updates

             

            Data Extract Tips and Tricks

            Tableau Data Extracts - Tips, Tricks and Best Practices | Tableau Software

             

            Quick Start: Aggregate to Visible Dimensions

            Quick Start: Aggregated Extracts

             

            Best Practices for Designing Efficient Workbooks (Best document on designing for optimal performance)

            https://alanattableau.files.wordpress.com/2015/07/designing-efficient-workbooks-v92.pdf

             

             

            Please note this response, if you find it helpful and/or if it thoroughly answers your question.

             

            Steve Burger

            Tableau Online Deployment Advisor