Cloud warehouse migration for a carmaker

The client is a German carmaker running analytics on an ageing on-premises warehouse that struggled at model-launch peaks. Their analysts depend on the warehouse for the insight that drives commercial decisions.

The fixed-capacity system was both slow for analysts and expensive to run, paying for peak capacity year-round while still hitting limits when it mattered.

An ageing on-prem warehouse throttled analysts with slow queries and cost a fortune to run at model-launch peaks. Performance and cost were both working against the business.

Fixed capacity meant paying for the peak year-round while still hitting limits when it mattered. The economics of fixed hardware were the worst of both worlds.

The carmaker needed elastic performance and lower cost without any loss of trust in the numbers. Migrating could not introduce doubt about the data.

We migrated to an elastic cloud warehouse with dbt-managed models, so capacity scales with demand instead of a fixed ceiling. Elasticity fixes both the peak-time throttling and the year-round overspend.

We tuned the models and storage for cost as a first-class design goal, not an afterthought, so the savings were engineered in rather than hoped for. Cost efficiency was treated as a feature.

Before cutting over, we ran old and new in parallel and reconciled results to the penny, so analysts could trust the migration completely, and the cutover happened only once that parity held. Trust in the numbers was never in question.