Multiple Table Extracts in Tableau

So, you might have heard that we launched multiple table extract functionality in 2018.3

I wanted to share my experiences with the beta build so far. Who knows the actual build might make this experience even better 🙂


Data Setup

Let’s understand the tables first

mte_data_Setup

  • There are total of 2.7 million rows in the Security_SalesFact_Tall table (each row is an order)
    • Each order has a zip-code attribute and an employee attribute
  • There is an alignment table (Security_ZipEmp_Alignment)
    • that has emp to zip-code assignments (2445 rows)
  • The OrgMaster_Employees has list of all employees
    • more than 300k+
    • Out of these 300k+ rows, only 603 employees has order data
  • The Ref_ZipCodes has list of all zip-codes
    • more than 42k+
    • Out of 42k+ zip-codes, only 579 zip-codes have orders

Prior to this functionality

Extracted this data using the “Single table”

  • creates a total of 11.4 million rows
  • create a .hyper file of 275 MB
  • takes 500 seconds to create

With this functionality

Extracted this data using the “Multiple table”

  • creates a total of 3 million rows (you see that during extract creation, the count goes up to 2.7m (fact table), then starts from 0 and goes up to 300k (emp dim)… you get the picture)
    • when you do SUM([Number of Records]), you will see exact same number as in the denormalized extract (11.4 m)
  • create a .hyper file of 30M
  • takes 80 seconds to create

Quick Comparison

Faster extract creation – 84% reduction in this case

Smaller extract sizes – 90% reduction in this case

  • The queries are returning the same results
  • Examples below with

All Data

mte_all_data

With One State

mte_1_state

With One Employee

mte_1_employee
Summary

I am loving the first implementation of multiple tables so far, that exceeded the expectation I had of a beta release