Case study
Flipkart
E-commerce marketplace
Overview
Marketplaces combine B2C storefronts with seller tooling, logistics, and payments. Flipkart’s Big Billion Days style events are textbook cases of thundering herds on product detail pages and checkout.
Engineering focus: idempotent checkout, inventory reservations, and search stability when marketing drives unpredictable traffic shapes.
Technical problems at scale
Flash sale inventory
Limited-stock drops need atomic decrements, waitlists, and bot mitigation. Rate limits and proof-of-work style challenges appear at edge (WAF, CAPTCHA).
Catalogue and search consistency
Millions of SKUs with frequent price updates. Search indexes lag behind OLTP; defining acceptable staleness per surface (search vs PDP) is a product decision.
Returns, refunds, and state machines
Order lifecycle spans warehouses, carriers, and customer support. Explicit state machines and sagas avoid double refunds and lost packages.
Systems & patterns you will hear about
- OLTP + search indexes
- Cart & checkout services
- WAF / bot defense
- Warehouse management integrations
Case-study angles
Sketch the checkout saga: payment authorized but warehouse allocation fails—what compensating transactions run?
Why might showing cached prices on listing pages but fresh prices on PDP reduce load yet annoy users?