How Betclic Uses Its Datalake to Generate ... ?
summary
Betclic ingests Datadog logs and Snowflake events into S3, then uses AWS Glue and Step Functions to transform that data into Gatling feeders which replay real user sessions in order. The pipeline supports scalable scenarios (+50% to +100% load), uncovers contention and behavioral patterns (e.g., users switching to casino at half‑time), and typically reduces an extraction from ~100GB raw to ~35GB processed with 30–60 minutes of end‑to‑end preprocessing while Gatling cloud nodes are scaled to the needed injection rate.
AI-written summary · about
chapters (12)
- 0:00Introduction
- 0:41Agenda
- 1:07Context
- 2:54Match evening
- 7:58Traffic analysis and reproduction
- 8:14User session concept
- 9:23Traffic retrieval
- 13:01Orchestration
- 17:11Pre-test process
- 19:22Feedback
- 25:31Extracting a day
- 26:59Technical blog
related talks
28mScalable AI & Web Insights with GitLab, Snowplow & Snowpipe
18mBehavioral Data Creation for AI - Delivering Value and Insights
10mBuilding a Data Platform at Assembly Payments
30mRaw Data to Business Insights: What You Need to Build a Modern Data Lake
23mDelivering Retail Analytics & Data Projects With Data Created For Purpose
01h 01mData Infrastructure in a Multi-Cloud Environment
24mGive meaning to 100 billion analytics events a day, analytics at Teads
24mForensic Log Analysis with BigQuery