twinalps
Book a call
Case 01 / anonymized engagement

A six-hour batch became a 22-minute run.

SnowflakedbtAirflowTerraform
6h22m
Engagement
14 weeks
Team
2 data engineers + Inpro lead
Environment
Financial services
Nightly batch runtimeNightly batch runtime across the engagement / minutes
6h3h22mW0W2W4W6W8W10W12W14Week 0 / on-premWeek 14 / Snowflake

Reporting stayed available on the legacy path until the new route was reconciled.

In detailLatency distribution after the work
fasterslowerp95 in target
Fast bins
Most requests land well under target
p95
Inside the agreed budget
Tail
Short - no long slow tail
The brief

Modernize the warehouse without interrupting reporting.

The existing on-prem warehouse ran a six-hour nightly batch. The migration had to move the data platform to Snowflake while keeping reporting live through the cutover.

Starting point6-hour batch
ConstraintParallel validation
ProofReporting stays live
Delivery

A production migration, owned in the client stack.

The delivery combined Snowflake, dbt, Airflow and Terraform. Two data engineers worked with an Inpro lead across the migration and cutover.

  1. SourceOn-prem warehouse
  2. OrchestrateAirflow
  3. Transformdbt
  4. ServeSnowflake + Terraform

The reporting path stayed live through cutover. Two data engineers worked with an Inpro lead for 14 weeks.

Engagement detail

The constraint, the shape and the stack in one view.

Constraint

Live reporting

The warehouse migration had to preserve the reporting path through the cutover.

Operating stack

Snowflake, dbt and Airflow

Transformations, orchestration and infrastructure were delivered in the production stack.

Delivery shape

14 weeks, senior-led

Two data engineers worked with an Inpro lead across the migration and handover.

Technical + business value

How the delivery translated into operating value.

Technical delivery

A controlled path from the legacy warehouse to Snowflake.

The implementation joined orchestration, transformation and infrastructure into one production migration. Airflow coordinated the jobs, dbt owned the transformation layer, Snowflake became the target platform and Terraform made the infrastructure repeatable.

Business value

A shorter nightly window without a reporting interruption.

The operational outcome was concrete: the same nightly workload moved from six hours to 22 minutes while reporting remained available during the migration.

22mNightly batch runtime after migration
LiveReporting continuity through cutover
Outcome

Reporting stayed live. The nightly runtime fell to 22 minutes.

6h22m
Your data platform

Bring us the bottleneck. We will bring the named engineers.

Start a brief
See other cases