Google SRE Prodcast

Rethinking SLOs with Narayan Desai

Apr 26, 2022
Ask episode
AI Snips
Chapters
Transcript
Episode notes
INSIGHT

SLOs Fit B2C Better Than B2B

  • SLOs work well for aggregate B2C systems but struggle for B2B clouds where every customer matters.
  • Aggregating diverse customer experiences into one SLO hides important per-customer differences.
ADVICE

Be Specific And Revalidate SLOs

  • Be skeptical of blanket SLO use and scope SLOs narrowly to answer one question.
  • Validate and iterate on SLOs over time so they reflect changing system behavior.
ANECDOTE

Quarter-Million SLO Proliferation

  • One internal Google service grew to about a quarter million parameterized SLOs, causing constant out-of-SLO states.
  • That scale made it unclear how to act and forced dilution of SLO usefulness.
Get the Snipd Podcast app to discover more snips from this episode
Get the app