The situation
- One Terraform mono repo had grown so large that plans timed out.
- The blast radius of each change was too large.
- Only one half-time engineer worked on DevOps.
What we did
Layered infrastructure as code
We wrote an implementation plan with layered Terraform modules for automation bootstrap, IAM, shared networking and services, and application landing zones.
EKS design and rollout
We designed, documented and implemented EKS, and completed the Kubernetes documentation.
Messaging architecture
We worked on the messaging architecture and gave the team a presentation on Kafka.
BI design recommendations
We recommended how to organize DynamoDB data in Snowflake, consolidate dimension data, and evaluate ETL and ELT paths back to DynamoDB.
Team support
We helped Nevro staff with Terraform and reported progress in bi-weekly delivery updates.
The result
- Nevro adopted the layered approach, and all new projects would use the multi-level architecture.
- The team implemented a Snowflake and Lambda initiative.
- Nevro approved our recommendations to add a Kafka and AWS architect and to assess its Snowflake environment.