Smarter Workflows, Leaner Costs: Inside Yum! Brands' Temporal Evolution
summary
Yum! Brands runs ~40 million Temporal workflows per month (about a billion actions) and reduces cloud spend by using Temporal Workflow Update to eliminate a reply activity (cutting ~20% of activity costs), switching event emission to local activities to collapse multiple actions into one, adding updatable timers to shrink retained storage and speed payments, and enabling codec servers for payload compression; combined changes yield roughly 12% lower Temporal cost plus fewer DB round trips and simpler code.
AI-written summary · about
chapters (16)
- 0:00Introduction and welcome
- 1:07Yum Brands and Bite Commerce
- 2:33Order lifecycle and challenges
- 5:44High-volume workflow characteristics
- 6:50Understanding Temporal costs
- 9:18Status updates and polling issues
- 11:05Synchronous response with workflow updates
- 13:11Cost savings from workflow updates
- 14:35Local activities for event publishing
- 16:21Problems with long-running timers
- 17:47Updatable timers for efficiency
- 18:55Codec servers and payload compression
- 19:56Reducing database transactions
- 21:00Results and cost savings summary
- 21:43Key takeaways and workflow design impact
- 22:47Future plans and next steps
related talks
15mScaling AI Agents Without Breaking Reliability
28mSurviving migration madness - Humana's move from Terraform Enterprise to Terraform Cloud
31mDOES14 - Shakeel Sorathia - Ticketmaster
41mMore Bang for Your Buck: How Yelp Autoscales Mesos + Marathon on AWS Spot Fleet
37mBeyond Buzzwords: Riding the Wave to Turn Trends into Profit - Duncan Winn, Keith Strini ,
36mPractical Cloud Native: What Works, What Doesn't
31mProduct Q&A
26mTransformation From a Traditional Organization to Embracing Agility and DevO