Choosing Reliability Targets and Their Cost

#96 - Practical Guide to Implementing SRE and SLOs - Alex Hidalgo

Topic 4 of 11

Published Jul 11, 2022 · Clip 27:14–34:29 (7:15)

Alex explains that how reliable a service should be depends on the service and its users — there's no formula, so be thoughtful and meaningful, talk to users, consult product teams and user journey documents. He then details why higher reliability targets are exponentially more expensive: redundancy multiplies components, more engineers are needed, on-call must be follow-the-sun across geographies, and multi-cloud setups add complexity. He notes four nines leaves only seconds of downtime per month, and that leadership often picks nines because Google does — "you're probably not Google."