Madrid devops forward devops is not what you think
Download as PPTX, PDF0 likes126 views
The document discusses the evolving concept of DevOps, emphasizing its role in enhancing agile software development by promoting collaboration, continuous delivery, and a culture of experimentation. Key practices include developers on-call, infrastructure automation, and designing for failure, all aimed at optimizing the value stream within technology companies. It concludes that DevOps is fundamentally a mindset that incorporates principles of agile development and continuous learning.
Madrid devops forward devops is not what you think
- 1. Devops is not what you think Madrid DevOps - Forward March 2019
- 2. 2
- 3. @eferro Eduardo Ferro Aldama https://www.eferro.net Sysadmin, Developer, Product, Tech lead, XP/Devops advocate
- 4. Definition
- 5. “Devops is... Cool name for sysadmins 5 A team that deploys other teams services A team that operates other teams services
- 6. “Devops is... learning to work in small batches 6 creating fast feedback loops enabling cross-functional collaboration though value streams developing a culture of experimentation” Jez Humble
- 7. DevOps includes and enhance the ideas of the Agile Software Development giving a more end to end vision of the value stream for a technology-based company. And, you know, Every Company Is A Tech Company. 7 @eferro
- 9. 9
- 10. 10 Biz/Prod Dev Ops AGILE SOFTWARE DELIVERY
- 11. 11 Devops (optimize whole value stream)
- 12. o Collaboration o Avoiding SILOs o Flow o Remove waste o Build Quality in o Create fast feedback loops o Maximize learning 12
- 14. o Developers on-call o Infrastructure Automation o Self Service Platform o Design for failure o Continuous Delivery o ... 14
- 15. o Pain Driven Development o Ops requirements in backlog o Production ready code o Instrumentation o Monitoring 15 Developers on call
- 16. “You build it, you run it. 16 Werner Vogels CTO Amazon
- 17. 17 Developers on call Practices & patterns Actionable Alarms Push metrics Pull metrics Centralized logging Observability Design for Failure Tools Kibana Prometheus Grafana Fluentd Logstash PagerDuty
- 18. 18 Infrastructure Automation o Infrastructure as code o Configuration Management
- 19. Tools Puppet Chef Ansible Terraform Consul Docker 19 Infrastructure Automation Practices & patterns Infrastructure testing Inmutable Infra Image bakering Containers Push conf management Pull conf management
- 20. o Each team o Deploy by themselves o Operate their services 20 Self Service Platform
- 21. 21 Self Service Platform Practices & patterns Deploy pipelines One command deploy Blue Green Deployments Rolling Deployments Platform Control Panel Alerts Monitoring Tools Kubernetes OpenShift CloudFoundry Heroku Docker OpenStack Jenkins
- 22. o Embrace failure o Design for failure o Learn from it 22 Design for Failure
- 23. o Optimize recovery (MTTR) o Observability o Monitoring o Automatic recover 23 Design for Failure
- 24. 24 Design for Failure Practices & patterns Circuit Breakers Retries Blameless IRs Chaos Engineering Game days Testing in production Smoke tests Tools Hystrix HA Proxy Chaos Monkey Envoy OpenTracing
- 25. Build, Test, Deploy your apps in a fast and automated manner. 25 Continuous Delivery
- 26. o Continuous Integration o Deployment Pipelines o Deploy during working hours o Automatic testing o Release vs Deploy 26 Continuous Delivery
- 27. 27 Continuous Delivery Practices & patterns XP practices BDD Deployment pipelines Rolling Deploys Feature Toggles Parallel Changes Tools Cypress Jenkins CircleCI GitLab launchdarkly
- 29. 29
- 30. 30
- 31. 31 Self Service Platform Product Team A Product Team B Platform Team Production Engineering DevOps Team SRE Team
- 32. o Deployment Frequency o Lead time for changes o Time to restore o Change Failure Rate 32 Metrics
- 33. o Deploy during working hours o Self Service deployments o Blameless IR o Dashboard and Metrics 33 Inflection points
- 34. References
- 35. 35
- 37. 37 Conclusions Devops is a mindset Devops includes/requires Agile Development Continuous Delivery Flow, Feedback, Continuous Learning
- 38. “ To make error is human. To propagate error to all server in automatic way is #devops.” – DevOps Borat. 38
Editor's Notes
- #8: http://www.eferro.net/2018/05/devops-concepts-and-learning-path.html https://www.forbes.com/sites/oreillymedia/2014/05/13/every-company-is-a-tech-company/#47fc2c1420a0
- #10: Value Streams Customer / Ideas -> Working Software in Production Customers: -External customers -Internal customers (Financial Depto, Delivery Teams, Value Stream: are Business Outcome Driven Value Streams samples SaaS Product Value Stream: From external customer feedback to new feature in a SaaS product. (New feature) Internal IT value Stream: From internal customer request to a new internal ERP feature. (New feature, integration) Platform Value Stream: From internal customer (delivery team) to a new capability in our infrastructure or platform. (Optimized pipeline, New platform feature).
- #11: Walls Specialization Queues / Waiting time Waiting Misunderstanding Batch Oriented Huge lead time Huge cost of coordination Even Worst Sec Architecture QA Compliance No walls Agile Software Delivery Optimize Only Software creation A good improvement for late 90s early 2000 Packaged software Internal software 1 deploy / month Not Enough for: Age of software SaaS
- #12: No barriers / No walls Functional Optimization vs Business Outcomes
- #13: Main characteristics of DevOps culture: Collaboration between development and operations (avoiding SILOs and conflicting goals). Organize around the value streams optimizing for the flow efficiency (not resource efficiency). Learning to work in small batches. Remove waste (Non-Value Adding Activities). Of course, we talk about value from the customer point of view. Build quality in. Create fast feedback loops. Maximize organizational learning (making safe to fail and learn). Iterative Incremental Continuous Automated Self-Service Collaborative Holistic
- #21: Requires Code pipelines No hands-off
- #32: Several teams one for each value stream One of the value streams have the other teams as clients and are in charge of the platform…. But they are also a cross-functional team including all the roles. They don’t deploy other teams artifacts. The platform should be self service Good option when our previous structure have a infrastructure/systems team 1- Blameless IRs 2- Feedback 3- Build self service tools 4- Each product team deploys 5- Each product team on-call for their services