1 of 1 people found this helpful
please check the below links.
there are a few main factors when evaluating when to use extracts of live connections
#1 if security is of primary concern you may want to use a live connection. when extracting data you will lose your database authentications and security. while you can rebuild security models in tableau through user filters and other joins this may be cumbersome.
#2 if you are connecting to a slow performing database and query execution time is poor, you may want to utilize an extract approach
#3 with Tableau's current extract engine, the number of records in your extract should not exceed 300M or so. this will be changed hyper in the next few releases. in the event where you have data that exceeds this size, we recommend a few approaches identified below
a) consider using live connections against a fast analytics database like SAP HANA, Google Big Query, or Redshift.
b) consider using aggregated extracts - sure your raw data may have more than 300M records but you NEVER show that many in Tableau. so aggregate the data prior to building an extract to get good performance in tableau.
In which scenarios extract will give worst performance ? is it if it cross 300M rows only ? or elase any other scenarios?
300M is an ambiguous number - but it is around there that we have seen users experience degraded performance. if you have a smaller data set and are experiencing poor performance it is likely due to the dashboard design and layout.
i tell my customers that building a dashboard is similar to building a budget. you only have $100 to spend, so spend it wisely. everything that you render on a sheet & dashboard costs something.