Zalando Tech: From Java to Scala in Less Than Three Months
10 likes7,818 views
The document discusses Zalando's technological transformation, highlighting the migration from a legacy tech stack to a modern architecture utilizing Scala and Play Framework. It provides insights into the challenges faced during this transition, the reasons for choosing Scala, and the company's commitment to fostering a culture of learning and collaboration in software development. Key aspects include the adoption of microservices, the importance of functional programming, and the development of internal training for staff to enhance their skills.
Zalando Tech: From Java to Scala in Less Than Three Months
- 1. From Java to Scala in Less Than Three Months
- 2. Alexander Kops Started at Zalando in 2010 Delivery Lead, Brand Solutions
- 3. 15 countries 3 fulfillment centers 16+ million active customers 2.2+ billion ā¬ revenue 2014 130+ million visits per month 10.000+ employees, ~900 in tech Zalando at a Glance Visit us: tech.zalando.com
- 4. Legacy Tech Stack 2008 - 2010
- 5. Legacy Tech Stack Complete rebuild in 2010
- 8. Rules of P lay Microservices REST APIs API First SaaS principle Cloud: Build on top of AWS
- 9. Brand Solutions Formed in 2014 Creating a CMS for our fashion brands Uploading their own content to the Zalando shop Five teams, 50+ people
- 10. My Teamās Challenge Build new analytics product Gather + compute data from different sources Potential future growth Work with legacy applications Be ready for cloud deployment
- 11. Why we chose Scala Experience with the JVM Works well with Java libraries Mature ecosystem Backed by Typesafe Functional programming good for parallelization
- 12. Why we chose Play Intuitive to use Excellent performance Extensibility Backed by Typesafe
- 13. Scala: the Good Parts
- 14. Concise
- 15. Combines FP and OOP
- 18. Great DSLs for testing
- 19. Scala: the bad parts
- 20. Compilation times Advanced language features add to compile time scalac -Xshow-phases But sbtās incremental compilation helps!
- 21. Language problems The same thing can be written in many different ways No canonical style guide Allows āoperator overloadingā FP code can get unreadable
- 23. Steep learning curve New language Functional programming concepts New frameworks New build tool
- 24. Helping the transition Assume 3-4 weeks overhead Two weeks exclusively for learning Encourage study groups Bring in freelancers Incremental learning OO -> FP
- 25. Provide material
- 26. Fitting projects Prototypes Greenfield projects Modular projects: Layered approach
- 27. Two types of Scala converts Using Scala as Java Functional programming hipsters
- 28. About coding style Develop own or embrace existing style guide Pair programming and code reviews Strict compiler settings
- 29. Tooling IntelliJ > Eclipse IDE sbt provides benefits over Maven and Gradle
- 30. Hiring Scala developers Genuine Scala developers are rare Senior Java developers eager to learn Scala Provide internal trainings Spread the word that you are doing Scala
- 31. Our application
- 32. Typesafe Stack in Zalando 40% of engineers want to learn Scala >12 teams are developing in Play and Scala >40 engineers Four Typesafe-certified Scala and Play trainers in-house
- 33. Other applications Merchant Center Fraud analysis Stylist website A/B testing tool Payment services
- 34. Play-Swagger āAPI Firstā library for Play Developed in-house by two engineers Collaboration with Typesafeās James Roper https://github.com/zalando/play-swagger
- 35. Follow us Tech Blog: tech.zalando.com GitHub: github.com/zalando Twitter: @ZalandoTech Instagram: zalandotech