Big Data at TUI

29.04.2019

Big Data has also found its way into TUI – and that in many ways! The classic 3 Big Data V's - Volume, Variety and Velocity - can be found here again:

Volume - large amounts of data
With over 90,000 hotels, 365 arrival days per season, more than 800 destinations and its own pricing challenge for every room, day and occupancy combination, it quickly becomes clear that really large amounts of data are coming together here.
Many optimization problems at TUI require the use of big data technologies in order to be able to analyze even large amounts of data quickly, cost-effectively and efficiently and make meaningful decisions based on them.
In addition to inventory data, we analyze transaction data such as click streams from website visitors and technical log messages from our aircraft fleet. Here, too, enormous amounts of data accumulate that would not be usefully processed without Big Data technology.
 
Variety - variety of data
Here, too, TUI has to deal with a large variety of data and unstructured data.
From hotel images, natural language hotel ratings or unstructured log messages - our data world is as diverse as the TUI product portfolio.

Velocity - Speed
TUI also uses streaming technologies in the Big Data environment. This allows us to go beyond the classic batch processing of data and implement modern lambda architectures so that we always have our finger on the pulse of the times.
In this way we ensure that today's holiday decisions are also based on today's data - quite up-to-date!

One step further: Value - business value
Another important V in addition to the classic 3 Big Data V's for TUI is the V for 'Business Value'. Big Data is not an end in itself at TUI, but every solution always has a clearly defined business goal in its sights.

How do we implement it?
To process the various tasks, we operate a Cloudera Hadoopcluster in the Amazon Cloud, among other things. By using the Cloudera Director, we can fall back on a highly developed technology from our partner Cloudera, which automates the setup of the cluster in line with the times. Using the Cloudera Manager, we can always optimally control all cluster resources and maintain an overview. Thanks to the integration into our company-wide "UC4 Automic" scheduling solution, the processes of the Hadoop cluster fit seamlessly into all other operational processes. Big Data and classic technologies work hand in hand.

 

Carsten Arndt (Senior IT Developer) & Johanna Schnier (Data Scientist)