SQLite Cloud and the AWS journey (part 1)

Why AWS?

AWS currently operates in over 40 regions worldwide, offering a strong global presence and unmatched infrastructure reliability.

When we began offering our SQLite cloud service, we initially launched in over 10 regions. However, to optimise service quality and infrastructure management, we strategically consolidated to just four key regions:

• us-east-2

• us-west-2

• ap-southeast-1

• eu-central-1

This allowed us to maintain performance while improving cost efficiency and operational focus.

AWS Benefits & Services

One of AWS’s major advantages is the flexibility it provides between fixed and on-demand costs. We make extensive use of core services, including:

• EC2

• EBS

• EKS

• S3

• ALB

Infrastructure mangement evolution

In the early days, we managed our infrastructure using Pulumi. However, we noticed high costs due to idle EC2 instances hosting SQLite nodes with minimal or no traffic.

To address this, we migrated our legacy infrastructure to Kubernetes, now managed through Terraform. This shift allowed us to significantly reduce idle resources and gain better control over our deployments.

EKS proved to be an ideal platform for running both on-demand and persistent workloads. Its seamless integration with other AWS services like VPC, ALB, and EBS greatly enhanced our operational efficiency.

As a small team, speed and simplicity are crucial. Through this evolution, we've cut deployment and provisioning times by over 50%.

Account design

The move away from our EC2 + ECS-based legacy setup gave us the perfect opportunity to redesign everything from scratch.

And it was the right move.

We now have a clean and robust account structure, with complete isolation between development and production environments right down to the domain level. Our new production account is lean, cost-controlled, and built for scalability.

We also introduced auto-scaling components, which help us save thousands of dollars each month by shutting down idle installations.

Access management is now centralised through AWS SSO, ensuring every team member can securely access the infrastructure and stay informed about system activity.

Centralised observability

Thanks to our optimised costs, we've been able to invest in an on-premises observability platform based on the ELK stack.

OpenTelemetry agents collect metrics from our services and Kubernetes and send data to a central point where we monitor service health and trigger alerts as needed. This gives us full visibility into system performance and helps us maintain reliability across the board.