Brightcart
Rebuilding checkout for Black Friday scale
A legacy monolith couldn't handle peak traffic. We migrated to a cloud-native microservices architecture on AWS with autoscaling baked in.
3×
Traffic capacity
0
Downtime incidents
180ms
Avg checkout latency
14 weeks
Engagement length
5 engineers
Team size
The challenge
Brightcart’s checkout flow ran on a five-year-old PHP monolith. Every November, traffic spiked 12× over baseline and the platform fell over within the first hour of Black Friday, costing an estimated $400K in lost sales the previous year.
Our approach
We decomposed the monolith into independently scalable services — catalog, cart, checkout, and payments — each deployed on AWS ECS with horizontal autoscaling tied to real-time load metrics. Checkout state moved to Redis to remove database contention at peak concurrency, and we ran three full-scale load tests simulating 15× traffic before launch.
The result
The new architecture handled Black Friday without a single downtime incident, sustaining 3× the prior year’s peak concurrent checkout sessions with sub-200ms response times throughout.
Built with
Have a similar problem to solve?
Tell us what you're working on and we'll scope it in a free 30-minute call.
Start your project