COPE with Node.js microservices

An overview of different approaches to content management and data architecture, focusing on the principle of COPE using Node.js microservices.

Transcript

1. BragaJS, September 2016
   COPE with Node.js
   microservices
   Eduardo Bouças
   eduardoboucas.com
   @eduardoboucas
   

   View Slide

2. Part I — Discovery
   

   View Slide

3. View Slide

4. Going mainstream
   CMS % of all websites
   WordPress 26.6
   Joomla 2.8
   Drupal 2.2
   W3Techs.com, 25 August 2016
   

   View Slide

5. Content
   Management
   System
   

   View Slide

6. Content
   Digital asset
   Management
   System
   

   View Slide

7. Content
   Digital asset
   User
   Management
   System
   

   View Slide

8. Content
   Digital asset
   User
   Layout
   Management
   System
   

   View Slide

9. View Slide

10. Traditional (monolithic) CMS
    Core
    Database
    Admin UI
    Templates
    Example: WordPress
    MySQL PHP app
    /wp-admin
    WP theme
    

    View Slide

11. Problem #1
    Long-term commitment to tech stack and CMS
    components
    

    View Slide

12. Problem #2
    How to integrate with external systems on different
    tech stacks?
    

    View Slide

13. Part II — Headless
    

    View Slide

14. Headless CMS
    Core
    Database
    Admin UI
    API
    Web server
    Templates
    Native app
    

    View Slide

15. ● Separation of concerns: content and layout are
    decoupled
    ● Any front-end can be plugged in — RESTful +
    everyone speaks JSON
    ● A single editing interface can be used to publish to
    multiple devices
    Benefits
    

    View Slide

16. ● Separates data from design
    ● Reduces editorial overhead
    ● Content re-usable and future-proof
    Create Once, Publish Everywhere
    COPE
    

    View Slide

17. View Slide

18. Problem #1
    Web CMS still pivot of entire system.
    

    View Slide

19. Problem #2
    Database, core and API still tightly coupled
    together. What happens if the publishing interface
    needs to change?
    

    View Slide

20. Problem #3
    COPE as an afterthought.
    

    View Slide

21. View Slide

22. ● Data architecture and business logic agnostic of
    tech stack
    ● Features as modular, interchangeable
    components
    ● Universal language
    ● Scalable on a micro-level
    A wish list
    

    View Slide

23. Part III — Microservices
    

    View Slide

24. «An approach to developing a single application as a
    suite of small services, each running in its own process
    and communicating with lightweight mechanisms, often
    an HTTP resource API.»
    Microservices
    Martin Fowler
    

    View Slide

25. API-first approach
    Modular components
    COPE by design
    Built in Node.js
    Scalable
    Free and open source
    

    View Slide

26. API
    A high-performance
    RESTful API layer
    designed in support
    of API-first
    development and the
    principles of COPE.
    Publish
    A writer’s window to
    the world of content
    creation. Flexible
    interfaces designed to
    optimise editorial
    workflow.
    CDN
    A just-in-time asset
    manipulation and
    delivery layer
    designed as a
    modern, global
    content distribution
    solution.
    Web
    A schemaless
    templating layer that
    can operate as a
    standalone platform
    or with API as a full
    stack web application.
    The suite
    

    View Slide

27. The DADI stack
    DADI API
    Database
    DADI Web Templates
    DADI CDN
    DADI Publish
    Media
    Storage
    Native app
    

    View Slide

28. DADI API
    DB shard
    iOS app
    DADI API
    DADI API
    Android app
    iOS app
    (CMS)
    DADI Web
    DADI Publish
    DADI CDN
    DADI CDN
    Real example
    DB shard
    DB
    cluster
    

    View Slide

29. ● > 100 million users a month
    ● Editorial teams of up to 400 people
    ● 250,000+ peak concurrent users
    Some numbers
    

    View Slide

30. ● Scalability
    ● Developer liberation
    ● Fast to market
    Benefits
    

    View Slide

31. Concurrent development
    API
    iOS Android
    Web
    CMS
    CDN
    iOS CMS
    Team 1
    Team 2
    Team 3
    Team 4
    

    View Slide

32. Track
    Real-time, streaming
    data layer providing
    accurate metrics at
    individual and
    product level. Drives
    DADI Machine.
    Visualise
    A data visualisation
    interface for DADI
    Track, but capable of
    taking data feeds
    from virtually any
    source.
    People
    CRM layer that
    guarantees
    uniqueness of
    individuals – and
    powers segmentation
    – in the anonymous
    and known space.
    Machine
    A machine learning
    layer that predicts
    user behaviour at an
    individual level based
    on past interactions.
    In the pipeline
    

    View Slide

33. Part IV — ?
    

    View Slide

34. Thank you.
    dadi.tech
    github.com/dadi
    

    View Slide