action alert allows application approach automation availability balancing better build canary cause Chapter complex components configuration consider continuous create customers dependencies deployment development team discussed documentation effectively effort engineers ensure error budget event example expected experience failed failure Figure goal Google identify impact implementation important improve incident increase infrastructure issues language load logs machines measure meet metrics mitigate monitoring move multiple on-call Once operational organization outage overload performance pipeline planning possible postmortem practices principles problems production recommend reduce release reliability requests response result scale server serving share single SLIs solution specific SRE team step strategy Study tasks testing tickets tion toil traffic types users