There are 3 different category of data: Hot, Warm, Cold data. 

Hot data is high level summary data. Hot data usually stored in Tableau Extract in-memory Hyper database, OLAP cubes, MPP databases, etc. 

Warm data is summary data stored inside Datawarehouse / Data marts (usually in a Star Schema design). 

Cold data is detailed raw transaction data stored inside the application OLTP databases or in Cloudera Data Lake (Hive / Impala). 

Tableau Extract is made for storing Hot / high level summary data (aggregated to day/month/year). And we also recommend our customer to create many small sizes Extract files (filter the data by City / Segment / Year / etc) for better performance.

 Tableau can help you to analyze Big data but there are certain best practices that you must follow, check out this excellent blog post from tableau here.  

Basically you need to put the right data in the right storage technology. So you can store high level “Hot” summary data inside Tableau Extract/Hyper, and you should have Datawarehouse to store “Warm” Star Schema summary data, and keep the detailed “Cold” raw transaction data in the application OLTP database or copy it over to Cloudera Data Lake using Hive / Impala. 

Then you can use Tableau to create:

1. High level summary dashboard querying “Hot” Tableau Extract data.
2. Summary dashboard “live” querying “Warm” Datawarehouse star schema summary data.
3. Detail transaction dashboard “live” querying “Cold” raw transaction detailed data stored inside OLTP database / Cloudera Impala.
4. Use Tableau Dashboard “Filter Action” to drill down from High level summary dashboard to Summary dashboard to Detail transaction dashboard. Filter action will do filtering when querying on the data, so it will only query the subset of the data following the items in the filter.

Separating your data into their appropriate storage following the Hot Warm Cold storage architecture and using Tableau “Filter Action” to drill down from Hot dashboard to Warm dashboard and to Cold dashboard should be much faster as compared to a single Dashboard showing all detailed transaction data using many filters for slicing / dicing the transaction data.

I made 3 samples video tutorial to show you how to create it using tableau:

Hot to Cold Drill Down Part 1 – create “Hot” high level summary extract data and publish it to Tableau Server

Hot to Cold Drill Down Part 2 – create “Hot” high level summary view by live connect to the “Hot” high level summary extract published data source

Hot to Cold Drill Down Part 3 – use Filter Actions to Drill down from “Hot” high level summary view to “Cold” Detail transaction view

Following the Hot Warm Cold data architecture best practices is the recommended strategy for every use cases. It can guarantee fast performance data analysis for everyone and very future proof.

Enjoy today.