How Open Lakehouse Architectures Drive Data and AI Platform Modernization

At GEODIS, we have launched a complete data and AI cloud platform modernization with the three main objectives. First, reduce Platform engineer workload with a range of best of breed managed services—including Databricks. Second, keep all the development and investment done on our previous architecture "compatible" with the new one—no bing bang. And last and most importantly, keep the platform fully open, interroperable and modular.In that context, we have been able to integrate with our legacy Cloudera technology in MS Azure new components—Astronomer, Privacera, Databricks and Starburst, relying on open standards and technologies—full Apache Iceberg™ data lake, catalog interroparability and Apache Spark™ data quality validation.I propose to discuss the challenges we have faced like Iceberg interroperability—Iceberg v2 vs v3, multi-engine integration—Databricks, Starbrust/Trino and Cloudera, and migration strategy—performance and interroperability enforcement.