Reactive programming and Hystrix fault tolerance

Reactive programming is a new paradigm that provides asynchronous event-based flow control. Java implementation is called rxJava and is being developed by Netfix. They have also released Hystrix - a non-functional layer that provides fault tolerance and latency features which also exposes reactive API.
At first, this talk will make you familiar with Reactive programming and introduce you to Hystrix. Then we will combine the two in effort of making a type-safe resilient Java API for REST services.

Transcript

1. Reactive programming and Hystrix fault tolerance Max Myslyvtsev
2. Agenda • Reactive • Observables • Hystrix • How we use it 2
3. Reactive Initiate Call Wait Get Response Continue 3
4. When call is complete Reactive Initiate Call Get Response Continue Continue 4
5. When call is complete Reactive Initiate Call Get Response React Continue 5 Observer pattern
6. Why Reactive? • Event-driven applications • Easy dependency composition • Allows parallel execution
 
 6
7. Why Reactive? • Event-driven applications • Easy dependency composition • Allows parallel execution • No callback hell 7
8. rxJava Observable<T> Future<T> Iterable<T> 8
9. Observable onNext onComplete onError 9 Terminal
10. Observable Observer onNext onComplete onError Subscription unsubscribe Observable subscribe Work start stop emit X
11. Operations • Create • Modify (immutable) • Terminate 10
12. Creation Create Observer onNext onComplete onError Subscription unsubscribe Observable subscribe onSubscribe X
13. Item Creation Never Just Empty Error Throwable 11
14. Creation From Iterable From Future Factory Defer Range 1 2 3 Start, Count 12
15. Transformation Map 13
16. Transformation FlatMap 14
17. Filtering Filter Take Skip 15
18. Combination Merge 16
19. Combination Concat 17
20. Combination Zip + 18
21. Aggregation 2 1 3 4 Reduce y x x+y + 10 19
22. Error Handling onErrorReturn 20
23. Error Handling onErrorResumeNext 21
24. Error Handling onErrorResumeNext 22
25. Source Observable Lift onNext onComplete onError New Observable onNext onComplete onError Operator 23
26. Termination BlockingObservable Single[OrDefault] First[OrDefault] Last[OrDefault] ToIterable ToBlocking Observable 24
27. Observable<List> Termination ToList BlockingObservable<List> Single[OrDefault] ToBlocking Observable 25
28. Temperature Cold • Lazy • Does not emit until subscribed • Whole sequence Hot • Eager • Emits regardless of subscriptions • Partial sequence 26
29. Scheduling Observable subscribeOn observeOn Schedulers Immediate Trampoline New Thread Computations IO From(Executor) Test 27
30. Immediate Unit of Work schedule Unit of Work Thread X
31. Trampoline Unit of Work schedule Unit of Work X Thread
32. New Thread Unit of Work schedule Unit of Work Thread 1 Thread 2 X
33. Computations / IO Unit of Work schedule Unit of Work Thread 1 Pool X
34. Test Scheduler Unit of Work schedule Unit of Work Thread 1 Virtual Thread X
35. Dependency Composition • Single Item Observables • Fork Join • If Then Else X
36. Fork Join Map Map Observable Observable Observable Observable Zip X
37. If Then Else Filter (If) Filter (Else) Observable Map Map Observable Merge X
38. Why Hystrix? • Latency • Fault-tolerance • Isolation 28
39. Command Hystrix Fragile Operation API Observable Future Sync Execute Remote Service Metrics Dashboard Configuration Fallback 29
40. How we use it 30 Server REST Resources Client Maven Plugin Generated Client Hystrix generate during build REST API RX API same API
41. References • http://reactivex.io/intro.html • https://github.com/ReactiveX/RxJava/wiki • https://gist.github.com/staltz/ 868e7e9bc2a7b8c1f754 • https://github.com/Netflix/Hystrix/wiki • https://github.com/mmyslyvtsev/rx-workshop 31