APPROACH
The Tredence team took a piece meal approach to deconstruct the problem:
- The team evaluated multiple Cloud-Data-warehouses on features such as cost, scalability, workload management
- The technology choice narrowed down to Snowflake after a month-long POT where AAS was replaced by Snowflake and performance was benchmarked by varying workloads and testing
- The findings were presented to the client technology council and was deliberated at length before the final decision was taken
KEY BENEFITS
- Snowflake was onboarded as the central CDW, features of Snowflake like, complete separation of compute from storage and auto-scalability based on varying workloads were instrumental in cost reduction
- Further advanced features like multi-clustered data-warehouses which enabled horizontal scalability improved performance
RESULTS
- Overall reduction in operational cost by ~35% for the same workload
- A slight improvement in performance due to compute horsepower and compression techniques available on snowflake
- Tredence developed its own proprietary snow-flake compute selection accelerator during the process