A platform team moved an engineer from the networking group to own the billing service. Two weeks later, a billing incident pages that engineer at night. They have never run the billing runbooks and do not know its dependencies. What does this illustrate, and what helps?
Show answer and explanation
Answer: C
The cold start problem in on-call describes a responder who lacks recent experience with a system and therefore lacks context during an incident. Remedies include shadowing experienced responders, keeping runbooks accurate, reviewing previous postmortems, and building a dependency map before the first real page. These steps create the working knowledge that is missing when someone rotates in. The term has nothing to do with machine boot times, cooling, or function execution.
Why the other options are wrong
- A. Cooling failures are physical facility incidents. They do not describe a responder who has rotated onto a service and lacks the context needed to handle a page.
- B. VM boot delay is a compute performance issue unrelated to on-call staffing. Prebaked images would not give the engineer any knowledge of billing dependencies.
- D. Serverless invocation latency describes how functions execute. Merging functions does not address the knowledge gap of a responder who has not worked on the service recently.