The client was running a high-traffic web platform on a single-AZ infrastructure that had experienced two significant outages in the preceding quarter. Their AWS spend was also significantly higher than it needed to be, with over-provisioned resources across most services. They needed a properly architected, fault-tolerant setup that could handle traffic spikes without downtime — and without continuing to overpay.
The approach
Starting with a full audit of the existing infrastructure, I documented every resource, identified the cost drivers, and mapped the failure points. The redesign followed AWS Well-Architected Framework principles: multi-AZ by default, auto-scaling at every layer, and infrastructure defined entirely in Terraform so nothing existed outside of version control.
Full AWS infrastructure audit — resource inventory, cost analysis, failure point mapping
Redesigned VPC with public/private subnet separation across 3 Availability Zones
ALB with WAF in front of Auto Scaling Groups — handles traffic spikes automatically
RDS Multi-AZ with automated backups and read replicas for reporting workloads
CloudFront CDN for static assets — reduced origin load by over 60%
All infrastructure written in Terraform — reproducible, version-controlled, reviewable
CloudWatch dashboards and alarms for every critical metric with PagerDuty escalation