Case studies / Data foundations
Data foundations · Grocery retail
Teradata to Snowflake, at grocery-retail scale
One of the largest grocery retailers in the US was running on an aging 180TB Teradata Enterprise warehouse. We partnered with their own engineering teams to migrate the whole warehouse to Snowflake on Azure — without disrupting a business of 1,500+ stores and 30 million-plus customers.
Versus Teradata, via Snowflake's pay-as-you-go model and elastic scaling.
On critical batch workloads, using dedicated warehouses.
The full warehouse of customer and product data, re-platformed end to end.
The challenge
An aging warehouse under a very large business
The client's Teradata Enterprise data warehouse had become a constraint: 180TB of customer and product data, growing roughly 10TB a year, on a platform that was expensive to run and hard to scale — for a business where reporting SLAs are not optional.
Before
- 180TB Teradata Enterprise warehouse, growing ~10TB/year
- High running cost, limited elastic scaling
- Performance pressure against reporting SLAs
- Large body of legacy code tied to the old platform
After
- Full warehouse migrated to Snowflake on Azure
- 30% lower cost via pay-as-you-go and elastic scaling
- Up to 40% faster queries on critical batch workloads
- Re-designed schemas, validated data, ready to scale
How we approached it
Analyse, migrate at scale, then tune for cost and speed
Redesign for Snowflake
We redesigned the database schemas for Snowflake instead of lifting the legacy Teradata structures across unchanged.
Automate conversion at scale
We developed the ETL with automated script generation to expedite legacy-code conversion across the warehouse.
Tune for cost and critical-workload speed
We tuned elasticity and dedicated warehouses over 10,000+ engineering hours and 24 months, delivered alongside the client's teams.
Under the hood
Next step
Carrying a legacy warehouse you've outgrown?
We re-platform large, business-critical data warehouses onto modern cloud foundations — cutting cost and lifting performance, without disrupting the business that runs on them.