Helping a Global Fashion Retailer Optimize Cloud
Data Analytics
Project Overview
Client Challenge: Redesign a critical, high-throughput service to be more scalable, reliable, and cost-efficient—while eliminating dependencies on deprecated systems and performing a no-downtime cutover. Solution: Analyzed the legacy OINK (Order-Invoice) service and designed OINK v2 with loose coupling and simplified upstream dependencies. Implemented a Kafka-first architecture with the right partitioning for pod scaling; added DLQ across components to avoid missed events. Ran v2 in production shadow mode; built a comparator to match v1 vs. v2 outputs; orchestrated safe traffic switch from v1 to v2 in collaboration with OMS/Payments/Tax/ERTM/Sales Audit teams. Results: ~4× lower cloud costs (3 pods in v2 ≈ 12 pods in v1). ~7× faster RTCIM event processing (with 12 pods). ~4× less memory at comparable scale. Successful zero-downtime replacement of the legacy service; improved stability and customer experience.Client
hello@agencyspotter.com417 12th St W Suite 110, Bradenton, FL 34205 USA© 2012-2026 Agency Spotter LLC