Issues with asynchronous interaction

Transcript

1. Issues with asynchronous
   interaction
   @gaar4ica

   Anna Shcherbinina
   

   View Slide

2. I like Ruby, Arch & Artec3D
   

   View Slide

3. View Slide

4. Downtime?
   Never heard of…
   

   View Slide

5. Service Interaction
   Patterns
   

   View Slide

6. Interaction Patterns Asynchronous Request-Reply
   with Timeout
   

   View Slide

7. Interaction Patterns Asynchronous Request-Reply
   with a Notification Timer
   

   View Slide

8. Interaction Patterns One Request, Multiple Responses
   

   View Slide

9. Interaction Patterns One Request, Mandatory Response,
   and Optional Response
   

   View Slide

10. Interaction Patterns Subscription
    

    View Slide

11. Interaction Patterns Unsolicited Notification
    

    View Slide

12. Scaling from old
    time to now-days
    

    View Slide

13. Be ready for capacity
    

    View Slide

14. Axes of scale
    Split by Function,
    Service or
    Resource
    Horizontal
    duplication
    Lookups or
    Formulaic Splits
    

    View Slide

15. • What: Typically called horizontal scale, this is
    the duplication of services or databases to
    spread transaction load.
    
    • Why: Allows for fast scale of transactions at
    the cost of duplicated data and functionality.
    
    X-axis
    Clone things
    

    View Slide

16. • What: Sometimes referred to as scale
    through services or resources, focuses on
    scaling data sets, transactions and
    engineering teams.
    
    • Why: Allows for efficient scaling of not only
    transactions, but very large data sets
    associated with those transactions.
    Y-axis
    Split different
    

    View Slide

17. • What: This is very often a split by some unique aspect of
    the customer such as customer ID, name, geography and
    so on.
    
    • Why: Rapid customer growth exceeds other forms of data
    growth or you have the need to perform fault isolation
    between certain customer groups as you scale.
    Z-axis
    Split similar
    

    View Slide

18. What prevents us from
    successful scaling?
    

    View Slide

19. Stateless - is the
    common guide
    

    View Slide

20. Be ready for failure
    

    View Slide

21. Failure
    Monitoring
    Monitoring:
    • All server event logs;
    • Monitoring for proper updates;
    • Backup status;
    • Firewall activity, hack/spam relay attempts;
    • System bottlenecks (system processes, processor/ram
    utilisation, etc.);
    • Application services;
    • Real-time Inventory collection and web-based reporting;
    • Remote support for most issues.
    

    View Slide

22. Failure
    Monitoring
    Monitoring:
    • All server events;
    • Backup status;
    • Firewall activity, hack/spam relay attempts;
    • System bottlenecks
    • Application services.
    

    View Slide

23. Failure
    Log aggregation
    How to understand the behaviour of an application and
    troubleshoot problems?
    • All Use a centralised logging service that aggregates logs
    from each service instance. The users can search and
    analyse the logs. They can configure alerts that are
    triggered when certain messages appear in the logs.
    

    View Slide

24. Failure
    Distributed tracing
    

    View Slide

25. Failure
    Exception tracking
    How to understand the behaviour of an application and
    troubleshoot problems?
    • Report all exceptions to a centralised exception tracking
    service that aggregates and tracks exceptions and
    notifies developers.
    

    View Slide

26. Failure
    Learn from failure
    Leverage every failure to learn and
    teach important lessons
    

    View Slide

27. Failure
    QA
    • QA does not increase quality of your system;
    • QA decrease risks;
    • Use QA team to learn from past mistakes;
    • QA helps to decrease a number of repeated defects.
    

    View Slide

28. Failure
    Backups
    • Check your backups twice;
    • Store your backups away of your application;
    • Create backup for your backup.
    

    View Slide

29. Failure
    Deployment
    

    View Slide

30. Failure
    Deployment
    • Set up your deployment process to be ready for rollback;
    • Avoid irreversible releases;
    • Monitoring and metrics helps to measure and analyse
    result of the release;
    • Set up partial deployment.
    

    View Slide

31. Autoscaling
    

    View Slide

32. Autoscaling
    Monitoring
    • Monitoring is the key feature of autoscaling;
    • Metrics should be re-estimated once a quoter;
    • Use complex metrics to avoid wasting resources.
    

    View Slide

33. Autoscaling
    Dashboard example
    

    View Slide

34. Autoscaling
    Scale down
    • Top and bottom limits are vital to avoid money loss;
    • Do not forget to empty resources your app does not need
    anymore.
    

    View Slide

35. Interaction
    SPoF
    Single points of failure
    

    View Slide

36. Interaction
    SPoF
    Evolving application in it’s
    natural way
    

    View Slide

37. Interaction
    Interaction patterns
    

    View Slide

38. Be competent
    

    View Slide

39. Thank you
    https://twitter.com/gaar4ica
    https://github.com/gaar4ica
    Scalability Rules: 50 Principles for Scaling Web Sites (Martin L. Abbott, Michael T.
    Fisher)
    Practical Object-Oriented Design in Ruby (Sandi Metz)
    Eloquent Ruby (Russ Olsen)
    Ruby Performance Optimization: Why Ruby is Slow, and How to Fix It (Alexander
    Dymo)
    Design Patterns in Ruby (Russ Olsen)
    Implementing SOA: Total Architecture in Practice By Paul C. Brown
    http://microservices.io/patterns/
    

    View Slide