Facing a 2027 deadline to decommission Zeppelin and transition analytics workloads to Databricks, Deutsche Börse developed a novel generative AI-powered approach to automate structurally complex notebook conversions, preserving critical business logic while reducing redevelopment time drastically.
- Hybrid AI approach automates structural migration, preserving complex business logic.
- Databricks Apps streamline deployment without additional infrastructure.
- Migration tool reduces notebook redevelopment from hours to under 20 minutes.
Infrastructure signal
Deutsche Börse’s migration reflects broader cloud infrastructure trends emphasizing unified platforms like Databricks for scalable analytics. The end-of-life of Zeppelin on Cloudera by 2027 precipitated the move, pushing workloads to cloud-native environments with improved orchestration, observability, and integration with Oracle and HDFS data sources.
The solution leverages Databricks Apps to implement the migration tooling, eliminating the need for separate runtime environments and simplifying deployment. This approach reduces operational overhead and supports a smooth, scalable transition by automating deterministic transformations and preserving notebook metadata in native Databricks format.
Developer impact
Developers and business analysts face significant challenges when migrating complex notebooks containing heterogeneous SQL, Python, custom interpreters, visualizations, and scheduling logic. Deutsche Börse’s approach separates purely structural conversion from logic translation, automating only the deterministic parts while using generative AI to accurately reconstruct business-specific code.
This hybrid workflow enhances developer productivity by cutting manual rewriting tasks from multiple hours per notebook to approximately 15–20 minutes. The migration app’s user-friendly shadcn UI integration with Databricks Apps improves access and usability, allowing business users to engage in migration with minimal friction while retaining confidence that critical logic remains intact.
What teams should watch
Teams planning large-scale analytics platform migrations should investigate hybrid automation frameworks combining rule-based structural conversion with generative AI for logic adaptation. This approach addresses the brittleness of pure rule systems and the complexity of diverse business-specific notebook content.
Observability and integration with enterprise data systems like Oracle and HDFS remain critical; the tool’s design deliberately leaves scheduling and native data references untouched to avoid disruption. Investing in extensible migration platforms that integrate smoothly with existing development workflows and cloud infrastructure can markedly reduce cost, error rates, and downtime.