From six hours to 22 minutes.

A warehouse migration to Snowflake shortened the nightly batch without interrupting reporting.

An anonymized client engagement.

Engagement: 14 weeks
Delivery team: 2 data engineers
with an Inpro technical lead

.renewed-site .page-index-mobile { display:none; } .renewed-site .page-index-desktop { display:flex; } @media(max-width:1000px) { .renewed-site .page-index-label { display:none; } .renewed-site .page-index-links { overflow-x:auto; flex-wrap:nowrap; justify-content:flex-start; width:100%; } }

Reporting could not pause for a migration.

The existing on-premise warehouse needed six hours to complete its nightly batch. Moving it to Snowflake meant changing the data platform while people still relied on its reports.

The constraint shaped the delivery: keep the legacy reporting path available until the new route had been reconciled.

Explore our data engineering work

Validate the new path before cutover.

The new platform ran alongside the existing warehouse. Reporting stayed on the legacy path until the new route was reconciled.

Keep reporting live
Reports keep reading from the existing warehouse while the new platform is built beside it.
Result No reporting interruption
Validate in parallel
Both paths run the same nightly workload until the results are reconciled.
Result Reconciled results on Snowflake
Switch the path
Reporting moves to the new route; the legacy path is retired once it is no longer needed.
Result A 22-minute nightly batch

How the migration kept reporting live

On-premise warehouse6 hours nightly

Same nightly workload, results reconciled

Snowflake22 minutes nightly

ReportsLive throughout

Step 1 of 3

Keep reporting live

Reports keep reading from the existing warehouse while the new platform is built beside it.

Validate in parallel

Both paths run the same nightly workload until the results are reconciled.

Switch the path

Reporting moves to the new route; the legacy path is retired once it is no longer needed.

Next: Validate in parallel ↓

Built in the client’s production stack.

Two data engineers and an Inpro lead worked across migration, cutover and handover. The warehouse, transformations, orchestration and infrastructure were delivered together.

dbt
The transformation layer. Data transformations managed as code within the production stack.
Airflow
The job orchestration. Coordinating the jobs that make up the nightly data workflow.
Snowflake
The target warehouse. The new platform for the nightly workload and the reporting it supports.
Terraform
The infrastructure. Repeatable infrastructure definitions for the new environment.