Retail, omnichannel, ~1,200 employees

Peak-trading auto-scaling for an omnichannel Irish retailer

Re-architected a retail e-commerce platform to handle a 40x peak traffic multiplier during sale events, reducing steady-state infrastructure cost by 38%.

The client ran a hybrid in-store and online retail operation with a Black Friday traffic multiplier that had historically caused outages. The existing cloud environment was sized for peak capacity year-round — expensive at baseline and still insufficient at the actual peak.

Our assessment identified four bottlenecks that prevented horizontal scaling: session state stored in-process, database connection pool limits, synchronous calls to a legacy inventory API, and absence of CDN configuration for static assets. We resolved each in a phased re-architecture.

With the architectural blockers resolved, we deployed auto-scaling groups for the application tier with policies calibrated to the client's observed traffic patterns. The first Black Friday post-migration saw a 41x peak traffic multiplier with zero application-layer errors.

Steady-state infrastructure cost fell by 38% from the pre-migration baseline, reflecting the elimination of over-provisioning for a peak that auto-scaling now handles dynamically.

Outcome

41x peak traffic multiplier handled without errors; 38% steady-state cost reduction; zero outage during post-migration peak trading periods.