Client Case Studies
MAPPY NATIXIS
Mappy, the French leader in local map search with 10 million unique visitors, produces more than 150 GB of raw logs daily on the 150 servers it has in operation.
A Hadoop platform was established to aggregate these logs, extract the relevant indicators and develop user sessions in order to analyse their behaviour.
The data is then enhanced by joins with some forty dimensions.
Problem
The various Mappy teams use Tableau directly connected to the Hadoop cluster to make the data easily accessible and usable by everyone.
Each analyst can thereby cross-reference all the data as they see fit without calling on technical services.
When users sent queries by Tableau dealing with billions of rows, they had a delay of several minutes for each interaction when using an engine such as Spark SQL or Impala.
With such long waiting times the data was not accessible in the true sense.
Solution
By connecting Tableau to Indexima, deployed on the Hadoop cluster in production, the response times fell from several minutes to around 100 milliseconds on average.
Navigation became fluid and instantaneous, even on dashboards that generate 25 simultaneous queries for each user click.