A Jedi's Guide to Migrating to Serverless

Transcript

1. A Jedi’s Guide to
   Migrating to Serverless
   CloudConf 2018
   Aleksandar Simovic
   Torino, Italia
   

   View Slide

2. What we’ll be covering
   • Stepping stones to serverless
   
   • Serverless monoliths
   
   • Rethinking application processes
   
   • The function-by-function approach

   “Divide and conquer”
   

   View Slide

3. STAR WARS
   NERD ALERT
   If you have Star Wars phobia or mania, take care!
   

   View Slide

4. Serverless means…
   • No longer need to manage or
   configure servers
   
   • Pay only for used time
   
   • Automatically scaled per usage
   
   • Auto-failover and automated
   high scalability
   

   View Slide

5. Sounds great!
   

   View Slide

6. How do we go serverless?
   

   View Slide

7. “We’ll just build applications on
   serverless instead of cloud”
   Initially serverless seems like…
   

   View Slide

8. Simply…
   Attribution: Simon Wardley
   

   View Slide

9. Also known as “The Luke Skywalker approach”
   

   View Slide

10. “We just don’t think about servers anymore.”
    “I have a light saber, I am a Jedi.”
    

    View Slide

11. ANTI PATTERN ALERT
    Welcome to Serverless Monoliths!
    

    View Slide

12. Serverless Monoliths = Dark Side
    • Hard to maintain
    and test
    
    • Slower starting time
    
    • Kept all drawbacks
    from your previous
    state
    
    • Higher cost
    

    View Slide

13. Bundling many functionalities into a
    serverless function also means…
    every time you wanted to use your car, 

    you would have to assemble it first.
    

    View Slide

14. “Now we just don’t think
    about servers anymore.”
    Serverless is much more than
    

    View Slide

15. Migrating to serverless requires us to
    change how we think about
    application development
    …not just jump to serverless.
    

    View Slide

16. So, how should we migrate?
    

    View Slide

17. Learn from Luke’s mistakes
    

    View Slide

18. Function-by-function approach
    “Divide and Conquer”
    1. Pick your already isolated services first

    “the low hanging fruit”
    

    View Slide

19. Function-by-function approach
    “Divide and Conquer”
    2. Rethink and separate services into functions
    Hint: you don’t pay for your function storage, just usage.
    

    View Slide

20. Function-by-function approach
    “Divide and Conquer”
    3. Divide your business processes into functions
    

    View Slide

21. Case Study:
    • Can’t predict the popularity of their pens
    
    • Have lots of unexpected peaks
    
    • Being able to serve is the key business
    goal
    
    • Spending lots on provisioning servers
    
    • Keeping lots of servers idle
    Problems:
    

    View Slide

22. Case Study:
    • Low hanging fruits 

    = preprocessors
    
    • Migrated all preprocessors into
    serverless functions
    
    • Separated their business process
    into functions
    
    • > 200,000 requests / h
    

    View Slide

23. Case Study:
    • Converters were eating too much
    resources
    
    • Too expensive Heroku dynos
    
    • Less isolation, could collide over
    resources
    
    • Hard to experiment
    Problems:
    MINDMUP
    

    View Slide

24. Case Study:
    • Low hanging fruits 

    = converters
    
    • Migrated all converters into functions
    
    • Separated business processes for payments,
    notifications, file storage
    
    • Half a million active users each month
    
    • Grew 30% in the migration period, 

    while reduced costs by 30%
    MINDMUP
    

    View Slide

25. Key points to remember
    • Identify your isolated problems.
    
    • Apply the “function-by-function” approach.
    
    • Analyze the remaining monolith services.
    
    • Repeat the approach.
    

    View Slide

26. “But can we use this approach
    everywhere?”
    

    View Slide

27. No. 

    Look at your current situation.
    

    View Slide

28. Case Study: Migrating User authentication
    

    View Slide

29. Reuse existing serverless
    components
    • Plug-and-play authentication and authorization
    services: AWS Cognito, Azure Active Directory, etc.
    
    • Easily add other authentication providers.
    
    • User groups.
    
    • Automated permission and policy systems.
    

    View Slide

30. Revised Steps
    • Identify isolated problems.
    
    • Check and replace these problems with existing
    serverless components.
    
    • Apply the “function-by-function” approach.
    
    • Analyze the remaining monolith services.
    
    • Repeat.
    

    View Slide

31. Summary
    • Serverless is much bigger than just 

    “no longer managing servers”.
    
    • Don’t just rush going serverless. Rethink.
    
    • Remember your context. Reuse.
    
    • Pick the low hanging “fruits” (services) first.
    
    • “Divide and conquer” your business processes into
    functions.
    

    View Slide

32. Learn more?
    Use code claudia40
    

    View Slide

33. Thank you!
    Questions?
    @simalexan
    

    View Slide