E-commerceCase study
Rebuilding a marketplace to survive festival-sale traffic
A multi-vendor fashion & lifestyle marketplace
The problem
The platform's single-server architecture went down during three consecutive festival sales, each time losing a full day of peak-season revenue. Average page load time during normal traffic was already a sluggish 4.8 seconds.
Our approach
- 01Split the monolith into catalog, cart, and order services so a spike in checkout traffic didn't take down product browsing.
- 02Moved product images to a CDN with WebP/AVIF delivery and added server-side rendering for product pages to cut load time.
- 03Added horizontal auto-scaling on the checkout service ahead of scheduled sale events, based on historical traffic data.
- 04Ran a full load test simulating 22x normal traffic before the next major sale.
Stack used
Next.jsNode.js microservicesRedisAWSCDN + image optimization
← All case studiesNext step
Facing a similar problem?
A 30-minute technical scoping call — no sales deck, just an engineer asking questions about your system.
Book a scoping call