DevOps incidents are up, but the bigger problem is believing platform resilience belongs to the vendor. Real resilience lives in governance, recovery design, and the engineering habits that survive a bad day.