Cloud expo 2018: From Apollo 13 to Google SRE - When DevOps meets SRE
3 likes405 views
This document discusses Site Reliability Engineering (SRE), which is Google's approach to service management. It outlines the key tenets of SRE, which include ensuring a durable focus on engineering, pursuing maximum change velocity without violating service-level objectives, monitoring, emergency response, change management, demand forecasting and capacity planning, provisioning, and efficiency and performance. The document also discusses best practices for incident management in SRE and how DevOps and SRE can be applied in the enterprise.
Cloud expo 2018: From Apollo 13 to Google SRE - When DevOps meets SRE
- 1. FROM APOLLO 13 TO GOOGLE SRE WHEN DEVOPS MET SRE Sanjeev Sharma @sd_architect | http://sdarchitect.blog
- 2. #WHOAMI • 20+ Years in Software Development and Delivery • Past: IBM Distinguished Engineer and CTO for DevOps Adoption • Now: Global Practice Lead for Data Transformation at Delphix • Author of two DevOps books: • DevOps For Dummies: https://ibm.biz/ BdsPMX • The DevOps Adoption Playbook: http:// amzn.to/2hH7rt2 • Blog: https://sdarchitect.blog • Tweets: @sd_architect
- 3. WHAT IS SRE? “SRE is what happens when you ask a software engineer to design an operations team. ” Betsy Beyer, Chris Jones, Jennifer Petoff, and Niall Richard Murphy “Site Reliability Engineering.” Site Reliability Engineering (SRE) : Google’s approach to Service Management
- 4. RELIABILITY: THE REAL AVAILABILITY NUMBERS! How much downtime does 5-nines 99.999% availability translate to? • Daily: 0.9s • Weekly: 6.0s • Monthly: 26.3s • Yearly: 5m 15.6s 4-nines or 99.99% translates to downtime of: • Daily: 8.6s • Weekly: 1m 0.5s • Monthly: 4m 23.0s • Yearly: 52m 35.7s Even the more common 99.95% availability SLO is a mere 43 seconds/day or 5:24 minutes/week.
- 5. APOLLO 13 – THE REAL HEROES Image Courtesy: Universal Pictures, NASA
- 6. EIGHT TENETS OF GOOGLE SRE 1. Ensuring a Durable Focus on Engineering 2. Pursuing Maximum ChangeVelocity WithoutViolating a Service’s SLO 3. Monitoring 4. Emergency Response 5. Change Management 6. Demand Forecasting and Capacity Planning 7. Provisioning 8. Efficiency and Performance
- 7. BEST PRACTICES OF INCIDENT MANAGEMENT 1. Prioritize 2. Prepare 3. Trust 4. Introspect 5. Consider alternatives 6. Practice 7. Change it around Image Courtesy: Universal Pictures, NASA
- 8. Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Test Stage Production Mainframe Hosted App Mobile App App Server Monolithic App Cloud Native App Enterprise Release Agile/Innovation Edge Rapid Delivery for Innovation • Agile • Antifragile • Experimentation • New and Innovative • Hybrid Cloud • IaaS/PaaS • Containers Industrialized Core Deliver at regular cadence • Agile • Stability • Predictability • Lean Delivery pipeline • Core and Legacy Systems Hybrid Infrastructure – Physical, Cloud • IaaS/PaaS • Containers Business Capability DevOps + SRE in the Enterprise
- 9. Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Development SCM Build Package Repo Deploy Test Stage Production Application N Application C Application B Application A Enterprise Release Agile/Innovation Edge Rapid Delivery for Innovation • Agile • Antifragile • Experimentation • New and Innovative • Hybrid Cloud • IaaS/PaaS • Containers Industrialized Core Deliver at regular cadence • Agile • Stability • Predictability • Lean Delivery pipeline • Core and Legacy Systems Hybrid Infrastructure – Physical, Cloud • IaaS/PaaS • Containers Business Capability Standardization Across Delivery Pipelines Deployment Automation and Orchestration Service andTest Environment Virtualization ArchitecturePlanning Release Management Operational Readiness
- 10. Your Delivery Pipeline will be as fast as the slowest Delivery Pipeline it is dependent on Data Friction is usually the last challenge to be addressed PLANNING
- 11. Modernizing to Microservices based Architecture: Refactoring Code and Data Code can be started afresh, not Data ARCHITECTURE
- 12. Developers are paid to write code, not maintain deployment and configuration scripts. DBAs are paid to Manage Data and Datastores, not generateTest Data sets APPLICATION DEPLOYMENT AND ENVIRONMENT ORCHESTRATION
- 13. If you are doing 2- week Sprints, but it takes 3-weeks to get aTest Environment andTest Data sets, how long are your Sprints? TEST SERVICE AND ENVIRONMENT VIRTUALIZATION
- 14. It is not possible to patch the software of a missile AFTER it has been launched RELEASE MANAGEMENT
- 15. Shift thinking from MeanTime Between Failure (MTBF) to MeanTimeTo Repair (MTTR). OPERATIONAL READINESS FOR SRE
- 16. ANTIFRAGILE SYSTEMS Antifragile: Things that are neither fragile or robust, but rather thrive in chaos.
- 17. WHEN DEVOPS MEETS SRE DevOps: “Everyone is responsible for delivering Business Value.” SRE: “(Everyone) is responsible for delivering Continuous Business Value”