Upgrade to Pro — share decks privately, control downloads, hide ads and more …

[2020.03 Meetup] [Talk #1] Mladen Milev - On th...

[2020.03 Meetup] [Talk #1] Mladen Milev - On the way to DevOps at Infinera

In this talk we will dive into the challenges a global telecom equipment provider encountered in their own R&D and operations divisions while embracing DevOps principles. We will follow-up on a project and its evolution from outsourced software developed under waterfall release cycle into an in-house developed agile and rapid release cycle project, bringing close development, tests and operation/customer.

Mladen Milev works as Internal Product Owner and Engineering Manager on an enterprise grade software solution for Infinera, a global telecom equipment provider. A defender of DevOps practices, he is helping in transforming traditional projects in DevOps enabled products.

DevOps Lisbon

March 30, 2020
Tweet

More Decks by DevOps Lisbon

Other Decks in Technology

Transcript

  1. On the way to DevOps at Infinera A successful Feature

    Development Model Mladen Milev DevOps Lisbon - 30 March 2020
  2. [email protected] @milev_mladen https://www.linkedin.com/in/mladenmilev/ About Me Mladen Milev • Software engineer

    with 17+ years experience • Lean Development and DevOps enthusiast • Author of a patent and patent pending applications 2 Hardware and Software Networking Solutions in 100+ countries across six continents
  3. The Product 4 Used by our sales and customer support

    (Internal Customers) Is used by selected end customers (External Customers) Enables new business and revenue
  4. Hands-On (Q2-2015) The Team • Product Manager – 0.2 •

    System Engineer – 1 • Developers – 2 • Quality Assurance – 2 5 310k Lines of Code 840 Days of SonarQube Technical Debt 25 Days Ready for Market Visual C++/win32
  5. A A A ABC ABC B B B C C

    C Release Execution (2015-2017) 7 Example release with 3 Features - A, B and C Specification Development Tests Integration Legend Regression
  6. SonarQube 8 72% decrease in Technical Debt over a 2-year

    period 0 100 200 300 400 500 600 700 800 G H I J K L M N O P Technical Debt (days) Q3/2015 Q3/2017
  7. Progress 9 0 50 000 100 000 150 000 200

    000 250 000 300 000 350 000 400 000 450 000 G H I J K L M N O P Lines of code 0 5 000 10 000 15 000 20 000 25 000 G H I J K L M N O P SonarQube Code functions Q3/2015 Q3/2017 Q3/2015 Q3/2017
  8. 1 4 8 5 8 7 8 7 3 1

    G H I J K L M N O P Number of Features in release Being Agile Agile practices but ... • The throughput was unstable • Some features were shifted to next • Quality issues, requiring maintenance release • Escalations from customer 10 Q3/2015 Q3/2017
  9. Team Perspective 11 Retrospective meetings indicated • Requests are high-level,

    • Hidden details • Difficult to elaborate accurate estimations • Late change requests
  10. Product Perspective 12 Product manager feedback: • Difficulty in planning

    releases • Shifting functionality to later release • Not fulfilling customer expectations
  11. 13

  12. Feature Development (Ownership) Ownership • Feature Responsible • Each feature

    should be user testable • Initial workshop • Feature Delivery workshop 15
  13. Feature Development (Process) Process • Each Feature ~ 1 week

    • Individual Work Package – 2/3 days • Limits are NOT strict but highly advisable 16
  14. Feature Development (Engagement) Customer Engagement • Workshops while developing •

    Participating in the acceptance tests • Presentation of delivered functionality from Dev to Customer 17
  15. FDD A A A A B B B C C

    C FDD B FDD C 18 3 Features 3 Features Traditional Approach New Approach Feature Development (Execution) ABC ABC Specification Development Tests Integration Legend Regression
  16. Feature Development (Execution) FDD A Legend FDD B FDD C

    New Approach 3 Features Reordering the execution is possible with no or minor impacts 19 Specification Development Tests Integration Regression
  17. Feature Development (Execution) Legend FDD D FDD E New Approach

    5 Features Linear scaling possible 20 Specification Development Tests Integration Regression FDD A FDD B FDD C
  18. How did it work? 21 25 2 0 5 10

    15 20 25 30 2015 2019 Minimum time to market (days) 14 2 0 2 4 6 8 10 12 14 16 2015 2019 Shifted functionality to later release
  19. 1 4 8 5 8 7 8 7 3 1

    2 4 7 10 14 15 G H I J K L M N O P Q R S T U V Number of Features in release 22 Results Once the process was established it was scalling-up Q3/2015 Q3/2017 Q2/2020
  20. Conclusions Achieved: • Decreased minimum time to market • Increased

    functionality per release • Less deviation from planned reoadmap Results in: • Increased stakeholders' confidence in the roadmap • Increased satisfaction of customer • Happier team 23