Analytics Databases: Unraveling the Secrets of Sisense Elasticube Design

Post on 15-Apr-2017

59 views 1 download

transcript

Unraveling the Secrets of Elasticube DesignInbar Rodan

● How many product units did we sell in total?

● How many employees are there per region?

● How many units did we sell in each product category?

● What Percentage of whole food Category are grain products?

Foodies Goodies

Sales Dashboard Creation

Business Question

Identify NeedsWhat Percentage of whole food Ctegory

are grain products?

KPI KPI KPI

Fact Dim Fact Dim Fact Dim

Import Data ScopeFocus Only on whole food

Dim Table

Dim Table

Dim Table

Dim TableDim Table

Dim Table

Dim Table

Dim Table

Transactions

Dim Table

Dim Table

Dim Table

Dim Table

Transform to a Centralized StructureTransform into product categroies

Dim TableDim Table

Dim Table

Fact Table

Connect Unique DimensionsRepresente to each category

Dim Table

Dim Table

Dim Table

Dim Table Dim TableFact Table

Validate the DataSo.. What is the grain food % ?

Validate

Best Practices Overview

Identify Import Model Connect

Excellent MistakesAND SECRETS

● How many product units did we sell in total?

● How many employees are there per region?

● How many units did we sell in each product category?

● What Percentage of whole food Category are grain products?

Foodies Goodies

Sales Dashboard Creation

Dim Table

Dim Table

Dim TableDim Table

Dim Table

Excellent Mistake #1“ Import all relationships and tables,

as in data warehouse “

Dim Table

Dim Table

Fact Table

Excellent Mistake #1Data Warehouse Implications

● Incorrect Results Query takes alternative path

● Long Import Times

1. Simpler Relationship2. Shorter Route3. Random

Path Preference

● Include only relevant tables, fields, relationships

● Consolidate by subject

Excellent Secret #1Solving Cycles

Excellent Mistake #2“ Let the Many-to-Many be! ”

Dim Table

Dim Table

Dim Table

Fact Table

● Incorrect Results How many units were sold per product?

● Performance Load Time Complexity of O(n2)

Excellent Mistake #2MtM Implications

Supplier Product ID

Pasta Buttini s.r.l. AA1

Tokyo Traders AA1

= 16Product ID Quantity

AA1 3

AA1 5

Product ID Quantity

AA1 3

AA1 5

Consolidate When tables data share granularity

Bridge When tables data does not share granularity, query from bridge table

Excellent Secret #2Solving MtM

OrdersSuppliers Per

Product

TableTable

Products

Suppliers/Orders

Excellent Secret #2Tracking MtM risks

1) MtM risk table in the Elasticube reflect the uniqueness of each table that has to be unique

2) MtM risk Widget

3) Add the KPI to Pulse

4) Let Pulse monitor your EC proactively

● Incorrect Results something is just wrong

● Multiple Dim Fields same dim in different tables

● Unclear Naming Convention of cubes, tables, fields

Excellent Mistake #3“ Once I finished modeling the

elasticube, it is done ”

● Verify Data

● Set duplicate fields as Invisible

● Pick Intuitive Naming Convention

Excellent Secret #3Solving Unfriendly Cube

Best Practices Summary

ValidateIdentify Import Model Connect

Questions?

Thank You!

Unraveling the Secrets of Elasticube DesignInbar Rodan