Resiliency in a service provider world
summary
Modern apps break on service provider failures so the concrete advice is to actually build redundancy not slogans. Avoid flip-the-registrar active standby name server switches because propagation makes outages worse. Prefer proven mechanisms like zone transfer when supported, or canonical config replication using tools such as OctoDNS or Terraform, or running providers in dedicated redundant networks. After last year’s large DDoS incidents redundant DNS setups prove practical and necessary.
AI-written summary · about
chapters (5)
related talks
51mDesigning for Failure: Build Resilient Services
25mAutomating Disaster Recovery: The Ultimate Reliability Challenge
41mHave You Tried Turning It off and Turning It on Again?
47mBuilding Resilience
48mResiliency and Availability Design Patterns for the Cloud
10mDesigning for Failure using AWS: Resilient Systems
22mResilience Is Not Enough
45m⚡A Resilient Application in a Partially Degraded World ⚡