5 Replies Latest reply on Mar 7, 2013 8:47 AM by Joshua Milligan

    Performance for large datasets

    Jeff Pressman

      I am researching a potential Tableau implementation that would include MS SQL Server as the primary data source for multiple tenants. With data for each tenant potentially in the hundreds of millions of rows I am curious what others in the community have found regarding performance.


      Are you pulling the data directly from the SQL engine or storing the data in an Extract?


      If you are using extracts are you hitting any significant performance walls as the size of your dataset grows?


      Has distributing the worker processes to different machines helped where performance issues arise?


      I'm sure 'it depends'...

      • On how wide the data is (ours will generally speaking be narrow)
      • On how complex the views are (understood)
      • On concurrent users (at this point data size is a greater concern for us than user base)
      • On your data structure and indexes (if going direct to the sql engine)


      Are there other things that impact performance that I should consider? What are your experiences?


      Thanks in advance!