All Day DevOps 2020 - Site Reliability Engineering

Technical debt as Pollution

• Replace the Technical Debt and its negative connotations with the concept of Continuous Solution Health
• There are ~∞ number of root causes for technical debt
• Far cheaper to prevent than clean up or mitigate
• Monitoring is required to ensure compliance
• Accurate accounting is essential
• Technical debt often imposes externalities on third-parties
• Slides

10 Steps to implement SLO in your Organization

• Slides

Site Reliability Engineering: Anti-patterns in Everyday Life and What They Teach Us

• Let’s start with the basic patterns :
  • Users should never notice an outage before you do
  • Engineer solutions to eliminate classes of errors rather that being satisfied with point fixes
  • Don’t feed the machines with human toil
  • Failure is an opportunity to improve, not to brandish pitchforks
• Slides

Driving Digital Transformation Through CloudOps and Site Reliability Engineering

• Some thoughts on team topologies
• Liked the progressive and evolving organization over Cloud Operations
  • Mode 1 : Traditional Ops
  • Mode 2 : Distributed Ops
  • Mode 3 : Decentralised Ops
• The so-called four pillars of CloudOps :
  • SRE
  • FinOps
  • GitOps
  • DevSecOps
• How to get organised and get started
• Measurement is key
  • Differend things to measure depending on if we’re talking about infrastructure/platform or product/application
• Get the basics right ; start slow but with confidence !
• Slides

Cloud as a Commodity: The highest probability of 100% availability

• Interesting antipatterns to resilient architecture
• Some takeaways :
  • “Human errors cause the bigger events”
  • “Environmental events cause AZ outages”
  • “Multi-Cloud with the CNCF can be efficient and minimal operational management overhead”
  • “Pipeline-as-Code ► Orchestration not Automation”
  • “Stop Making Infrastructure Pipelines”
  • “Include IAC / CAC in the delivery (environment delta)”
  • “Environmental requirements are part of the delivery (GitOps)”
  • “One pipeline to rulle them all”
• Have a look at Harness, formerly known as Drone one pipeline to rule them all
• Slides

Sampling in Open Source distributed tracing

• Interesting lecture about Observability and the problems around OpenTelemetry
• Slides

Back to All Day DevOps 2020 Recap