Deutsche Börse Group¶
Deutsche Börse Group (DBG) is a Frankfurt-headquartered financial-market-infrastructure operator. Its core businesses include the Frankfurt Stock Exchange (Xetra trading platform), Clearstream (post-trade settlement and custody), Eurex (derivatives exchange), and a range of data, risk, and benchmark services. As a market-infrastructure operator, DBG is regulated under European capital-markets law and operates with the operational reliability and audit posture appropriate to a systemically-important institution.
DBG appears in the wiki as a Databricks customer through a single 2026-05-19 customer-blog post: How Deutsche Börse built a generative AI tool to tackle the large-scale migration of Zeppelin notebooks to Databricks. Tier-3 single-article ingest.
What we know about DBG's data platform¶
The customer disclosure is narrowly scoped to the StatistiX platform — DBG's self-service analytics platform serving Clearing & Trading data. Disclosed properties:
- StatistiX serves ~95% of all Clearing and Trading data across the group.
- Hundreds of business users consume StatistiX directly.
- 2,000+ users are within scope of the upcoming platform migration.
- Legacy stack: Cloudera + Apache Zeppelin notebooks + HDFS + Oracle. Zeppelin notebooks are the user-facing analytics surface; HDFS and Oracle are the data backends.
- Forcing function: Cloudera is fully decommissioning Zeppelin in 2027. This is a vendor-end-of-life cliff, not a discretionary platform refresh.
- Target stack: Databricks (selected as the group's unified analytics platform). The platform decision is set; the open problem is the migration mechanics.
Other DBG businesses (Xetra, Clearstream, Eurex) are not disclosed in this source. The wiki record is limited to StatistiX.
What's disclosed about DBG's engineering posture¶
- Build-over-buy where the customer-specific work matters. DBG built the Zeppelin to Databricks Notebook Converter in-house rather than waiting for Databricks or a third party to ship it, on the rationale that "rewriting everything manually would take years" and existing rule-based migration tooling does not handle heterogeneous notebook content. The build is on the destination platform itself (Databricks Apps), so the migration tool runs inside the workspace it is migrating into.
- Joint engineering with the platform vendor. The post calls out "our collaboration with the Databricks team throughout the build helped us stay aligned and avoid rework."
- Disconfirmation of "agentic-first" design reflex. A first-attempt agentic architecture was discarded in favour of a simple UI + clean backend after the team determined the migration task was well-bounded enough that an autonomous agent loop "added overhead without solving the core problem."
- Self-service-for-business-users as a deployment goal. The migration tool is designed for non-engineers; the user workflow is six steps (export → upload → click Convert → download → upload to Databricks → paste prompt into Genie). "Migration doesn't require a dedicated engineering team."
DBG's role in the wiki¶
DBG is the customer-co-author of a Databricks-blog migration-architecture post that pins two reusable architectural primitives to the wiki:
- The structural-vs-logical-conversion split — separate the deterministic part of a heterogeneous-code migration from the non-deterministic part, and apply rules + LLMs to each respectively. Canonicalised at patterns/structural-deterministic-logical-llm-split.
- Context-encoded prompt handoff — the deterministic stage of a hybrid migration tool emits a prompt template populated with environment-specific facts that the user pastes into an LLM agent for the non-deterministic stage. Canonicalised at patterns/context-encoded-prompt-handoff.
DBG is also a canonical instance of a Cloudera-Zeppelin-to-Databricks migration motion that other large-enterprise Cloudera tenants will face by 2027.
Recent articles¶
- 2026-05-19 — How Deutsche Börse built a generative AI tool to tackle the large-scale migration of Zeppelin notebooks to Databricks — see sources/2026-05-19-databricks-deutsche-borse-zeppelin-to-databricks-notebook-migration. Customer-co-authored Databricks-blog post canonicalising the structural-vs-logical-conversion split (deterministic structural conversion in a Databricks App + heterogeneous logic reconstruction in Genie) for a 2,000-user, hours-to-minutes-per-notebook Cloudera Zeppelin migration. Forcing function: Cloudera Zeppelin 2027 EOL. New wiki primitives: systems/apache-zeppelin, systems/deutsche-borse-zeppelin-converter, concepts/notebook-format-migration, concepts/heterogeneous-code-migration, concepts/context-encoded-llm-prompt, patterns/structural-deterministic-logical-llm-split, patterns/context-encoded-prompt-handoff.
Related¶
- companies/databricks — vendor + co-author of the StatistiX migration writeup; destination platform.
- systems/apache-zeppelin — legacy notebook surface being migrated from.
- systems/deutsche-borse-zeppelin-converter — DBG-built migration tool.
- systems/databricks-apps — substrate the converter is deployed on.
- systems/databricks-genie — LLM agent that handles the logic-reconstruction stage of the migration.