action items alert anycast application automation autoscaling availability backend burn rate Chapter complex components config configuration language create customers dashboards datacenter debugging deploy DevOps documentation ensure error budget error rate evaluate event delivery Evernote example failure feature Figure frontend goal Google Google's identify impact implementation improve incident response infrastructure issues JSON Jsonnet Kubernetes latency line card load balancing LogJoiner machines Maglev MapReduce measure metrics microservices migration mitigate monitoring system multiple Niantic on-call engineer operational operational overload organization outage overload pager performance pipeline Pokémon postmortem practices prioritize problem product development team query release Reliability Engineering requests result Richard Murphy root cause scale server share SLIs solution SRE book SRE team Stephen Thorne strategy switch tasks team members testing tickets tion toil traffic trigger users VALET workflow workload YAML