Example journey. This page shows how we run this type of migration. It is not a named client result. References from comparable projects are available on request.
This example journey shows how we would move a retail bank’s risk and regulatory reporting from an ageing Hadoop cluster to Databricks: inventory, Spark jobs moved largely as they are, Oozie and MapReduce rebuilt as pipelines, and every report reconciled before decommissioning.
| Type | Example migration journey |
|---|---|
| Industry | Banking |
| Migration | Legacy Hadoop |
| Technologies | Delta Lake, Unity Catalog, Lakeflow Jobs, Lakeflow Declarative Pipelines |
Starting point
A bank runs risk and regulatory reporting on an ageing on-premise Hadoop cluster. Hardware renewal is due, upgrades are risky, and the few people who understand the jobs are stretched.
How we would approach it
- Assess: inventory Hive tables, Spark and Oozie jobs, and Ranger policies. Flag unused data for retirement.
- Plan: design a medallion lakehouse on the bank's chosen cloud, with Unity Catalog policies mapped from Ranger.
- Convert: move Spark jobs largely as they are, rewrite MapReduce and Oozie logic as declarative pipelines and Lakeflow Jobs.
- Reconcile: compare every regulatory report against the Hadoop output before sign-off.
- Cut over: move reporting teams in waves, then decommission the cluster.
Value the bank should expect
- No hardware renewal and fewer platform upgrades.
- Lineage from each regulatory figure back to its source.
- A foundation ready for fraud, AML and customer 360 use cases.
What to measure
Infrastructure and licence cost before and after, report run times, number of manual steps removed, and time to deliver a new report.
