TL;DR
From a snowy retreat in Norway to preparing for a customer’s Super Bowl ad, Stefan and Jens explored what happens when federation meets massive scale. Their takeaway: cache warming and reliability planning are what keep systems standing when the traffic spike hits.
Retreats and culture
The episode opened on a lighter note: the team’s annual retreat in Alta, Norway. Flights arrived in batches, with some arriving late at night into the Arctic circle. Between husky sledding and snowmobile rides, the week was about more than fun — for Jens, retreats are about culture-building.
Why cache warming matters
The main technical focus was a customer preparing for a Super Bowl ad. Jens explained cache warming: routers need to precompute query plans so that when millions of visitors arrive at once, no one waits seconds for the first request. Without it, those first hits during scaling could stall.
If a query hits a router and it takes 10 seconds to plan, that’s unacceptable. Cache warming makes sure no user ever sees that.
The business cost of slow
Stefan illustrated the business side with Coinbase’s famous floating QR code ad: it drove so much traffic their site crashed. Even if a site doesn’t go down, Jens added, latency kills conversions in e-commerce. Every second means lost sales.
Latency means you lose sales. It’s as simple as that.
REST vs Federation in the spotlight
This led to a broader point: federation isn’t just a technical preference over REST, it’s about visibility. With REST, frontend teams may not even realize all the calls they’re making until runtime. With federation, complexity is explicit and can be tested, warmed, and planned ahead of time.
Preparing for the Super Bowl
Stefan and Jens described preparations as a “war room”—but Jens stressed the real work is in the year before: load tests, schema checks, and simulation of traffic. They even maintain a database of past queries to replay and stress-test routers, ensuring the system can handle real-world patterns. If you’re scrambling during the game, you’ve already lost.
If we have to do something in the war room, it’s too late.
Looking ahead
For WunderGraph, supporting a Super Bowl ad was both a stress test and an opportunity. The episode closed with the reminder that reliability doesn’t come from hype—it comes from year-round preparation, safety nets in composition and query planning, and never leaving performance to chance.
This episode was directed by Jacob Javor. Transcript lightly edited for clarity and flow.
