Yum! Brands engineering manager shares how their e-commerce order platform evolved its Temporal workflows to cut costs at scale (40M+ workflows/month). Key optimizations include: adopting Temporal's native Workflow Update primitive to replace a complex polling/response-activity pattern (saving ~20% activity costs), switching fire-and-forget Kafka event publishing to local activities (reducing 5 actions per order to 1), implementing updatable timers to shorten workflow lifetimes when orders progress faster than expected, and using codec servers for payload compression to reduce storage costs. Combined, these changes saved ~12% of total Temporal costs while also improving latency, developer experience, and customer support load.
•23m watch time
131 Impressions