Fast In-memory Analytics
for Retail Data with
Columnar Databases
Ernesto Arbitrio - Valerio Maggio
arbitrio | [email protected]
Florence April 6, 2017
Slide 2
Slide 2 text
Retail Data
• Overview of data we have
• granularity
• refresh/update rate
• Quantity and storage required (space)
• services developed around these data
Slide 3
Slide 3 text
“Materialized Views”
• Description of what they are (non-technical)
• Some examples of Analytics we do on this data
Slide 4
Slide 4 text
The Problem!
~1 TByte Data
We need OLAP
Performance:
75M rows -> 5hours
Slide 5
Slide 5 text
The Solution!
Use a Column-oriented Database
(i.e. Just swap Rows with Columns)
Chuck Norris Test
Passed!