Opening Scenario
The all-hands ends with the CTO saying "we need a real DR story." Last quarter the primary region had a 90-minute partial outage; the team improvised, customers complained, and the postmortem named the absence of a documented recovery plan as the root cause. You inherit the project: define RPO and RTO for the streaming platform, design the recovery topology, write the runbook, prove it with a game day. The exec sponsor wants "zero data loss and immediate failover." Engineering reality says neither of those is free. Before designing anything, you need to align stakeholders on what's actually achievable, what each level of guarantee costs, and how to operate a system you can credibly fail over.
Question Progression
1. What do RPO and RTO actually measure, and why do they constrain everything?
Question. The CTO wants "zero data loss." Why isn't that a starting point?
Reasoning. Two numbers govern…
In the full tutorial
- When to Reach for This