[2019.02 Meetup][TALK] DevOps Adoption at Mercedes-Benz.io

In this presentation, I will show you our journey to achieving DevOps culture in our company. Since the beginning, we wanted to be able to quickly start delivering customer value in self-independent teams without an initial complex setup. I will also focus on how we deal with a fast growing company facing the challenge of giving freedom/autonomy to our teams. The pains we had and the lessons learned will be covered as well.

I will also deep dive a little bit on our technical stack and show how a team can set up a new microservice from scratch and run it in production in a couple of minutes. This is possible with some libraries we made on top of our CICD and PaaS (Pivotal Cloud Foundry).

André Macedo is responsible for the overall Architecture and Technology at Mercedes-benz.io. He is currently working together with the teams to define the technical roadmap and optimizing the software development life cycle.

Transcript

1. Highly productive delivery culture for continuous delivery André Macedo Architecture and Technology DevOps Lisbon 11 February 2019
2. About me • André Macedo • 35 years • Started my journey where 48k was enough • First programming: TCL Script @ mIRC • Telco, Media and Automotive • Architecture and Technology @ Mercedes-Benz.io Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 2
3. Mercedes-Benz.io • Born in 2017 • Driving the digital future of Mercedes-Benz • Offices in three locations (Stuttgart, Berlin and Lisbon) • Driven by 240+ employees + external partners • Delivering to 100+ countries Visit us at: https://www.mercedes-benz.io Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 3
4. Transition to Mercedes-Benz.io • Role “Transition Manager” • Plan for each component/project/team • Main Steps: • Analysis • Pairing • Target Architecture • Migration to our platform Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 4 What have we changed?
5. Optimizing Development Flow Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 5 Past Present ~ 30 days ~ 2-3 days
6. Organization What have we changed? • Holacracy instead of classical hierarchy • Organized in circles • People assigned to roles and accountabilities • No “top-down” decisions • Our individuals have: • Responsibility • Freedom • Autonomy • Fast decisions Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 6
7. Teams What have we changed? • Around 50+ teams • Responsible for the whole journey • Ownership (no need for any approval) • Focus on deliver features / business value • No handovers between teams • How can the teams have such independency? Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 7 Product Owner Team A Software Engineer Scrum Master UX QA Product Owner Team B Software Engineer Scrum Master UX QA
8. Platform How can the teams have such independency? • Give autonomy to the teams • Automate “almost” everything • Keep teams focused on delivering features • Reactive architecture, but decoupled components/services • Dedicated DevOps team Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 8
9. Platform How can the teams have such independency? Jenkins Pipeline Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 9 Database Creation on PCF Is it going well?
10. KPIs Is it going well? • Dashboard integrated with operational platforms: • Issue Tracking • CICD • Continuous Inspection • Monitoring Tool • Lead Time • Code Coverage • Technical Debt • Deployment Frequency • Average Release Duration Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 10
11. DevOps Journey Everything in place Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 11
12. Lessons Learned • Automation is never finished (ex.: build, deploy, databases, system access) • Decoupling of components and teams for parallel development • PaaS (PCF) is good to ramp up teams in a faster way, but adds a dependency from the team managing it • Change your feedback / exchange model when needed • Establish a collaboration model for some shared services (ex.: shared lib to deploy to PCF) • “Continuous Delivery – Sounds great but it won’t work here” – Jez Humble Highly productive delivery culture for continuous delivery | Architecture and Technology | 11/02/2019 12
13. Thank you Andre.Macedo@mercedes-benz.io https://www.mercedes-benz.io