Site Reliability Engineering: The DevOps Culture.
- 1. Site Reliability Engineering DevOps Culture and System Reliability Ahmed Badran - CTO @Magalix ahmed.badran@magalix.com
- 2. Agenda - My Background - Developers vs Operators - The DevOps Culture - Site Reliability Engineering - Service Level Objectives ○ Availability ○ Scalability - Resources - Q&A
- 4. Developers vs Operators With the rise of online systems and microservices architecture, agile development became the norm Agility vs Stability Applications vs Infrastructure
- 6. The DevOps Culture ❏ Reduce Organizational Silos ❏ Implement Gradual Change ❏ Accept Failures as Normal ❏ Leverage Tooling and Automation ❏ Measure Everything
- 8. Site Reliability Engineering ✓ Reduce Organizational Silos ➔ Shared ownership of production environment ✓ Accept Failures as Normal ➔ SLO & blameless postmortems ✓ Implement Gradual Change ➔ Agile development and small releases ➔ Canary testing in production ✓ Leverage Tooling and Automation ➔ Continuous integration and deployment ✓ Measure Everything ➔ Define and measure system reliability ➔ Measure and reduce toil
- 9. Service Level Objectives ● SLI Service Level Indicator Metrics over time which inform us about the health of a service ● SLO Service Level Objective Agreed about bounds regarding how often the SLI must be met ● SLA Service Level Agreement Business level agreement which define penalties if we fail to deliver the agreed on availability
- 10. Availability Good service to total demanded service ❏ Availability = Uptime / (Uptime + Downtime) ❏ Availability = Successful Requests / Total Requests
- 11. Availability
- 12. Availability
- 13. Scalability ● Measuring Load ○ Write load ○ Read load ● Measuring Performance ○ Throughput ○ Response time ○ Queuing time
- 14. Scalability
- 15. Resources ❏ Site Reliability Engineering - 2 books ❏ Building Microservices (Designing Fine Grained Systems) - Book ❏ Designing Data-Intensive Applications - Book ❏ On Designing and Deploying Internet Scale Services - James Hamilton 2007 ❏ Meaningful Availability - Google Paper ❏ 8 Fallacies of Distributed Computing - wikipedia
- 16. Q&A Ahmed Badran - CTO @Magalix ahmed.badran@magalix.com