Tweet
Tweet

Slide 1

Slide 1 text

Ben Christensen Developer – Edge Engineering at Netflix @benjchristensen http://techblog.netflix.com/ QConSF - November 2014 Reactive Programming with Rx

Slide 2

Slide 2 text

RxJava http://github.com/ReactiveX/RxJava http://reactivex.io Maven Central: 'io.reactivex:rxjava:1.0.+'

Slide 3

Slide 3 text

Iterable pull Observable push T next() throws Exception returns; onNext(T) onError(Exception) onCompleted()

Slide 4

Slide 4 text

Iterable pull Observable push T next() throws Exception returns; onNext(T) onError(Exception) onCompleted()  //  Iterable  or  Stream      //  that  contains  75  Strings    getDataFromLocalMemory()      .skip(10)      .limit(5)      .map(s  -­‐>  s  +  "_transformed")      .forEach(t  -­‐>  System.out.println("onNext  =>  "  +  t))    //  Observable      //  that  emits  75  Strings    getDataFromNetwork()      .skip(10)      .take(5)      .map(s  -­‐>  s  +  "_transformed")      .forEach(t  -­‐>  System.out.println("onNext  =>  "  +  t))  

Slide 5

Slide 5 text

Iterable pull Observable push T next() throws Exception returns; onNext(T) onError(Exception) onCompleted()  //  Observable      //  that  emits  75  Strings    getDataFromNetwork()      .skip(10)      .take(5)      .map(s  -­‐>  s  +  "_transformed")      .forEach(t  -­‐>  System.out.println("onNext  =>  "  +  t))    //  Iterable  or  Stream      //  that  contains  75  Strings    getDataFromLocalMemory()      .skip(10)      .limit(5)      .map(s  -­‐>  s  +  "_transformed")      .forEach(t  -­‐>  System.out.println("onNext  =>  "  +  t))  

Slide 6

Slide 6 text

Single Multiple Sync T getData() Iterable getData() Stream getData() Async Future getData() Observable getData()

Slide 7

Slide 7 text

Observable.create(subscriber -> { subscriber.onNext("Hello World!"); subscriber.onCompleted(); }).subscribe(System.out::println);

Slide 8

Slide 8 text

Observable.create(subscriber -> { subscriber.onNext("Hello"); subscriber.onNext("World!"); subscriber.onCompleted(); }).subscribe(System.out::println);

Slide 9

Slide 9 text

// shorten by using helper method Observable.just(“Hello”, “World!”) .subscribe(System.out::println);

Slide 10

Slide 10 text

// add onError and onComplete listeners Observable.just(“Hello”, “World!”) .subscribe(System.out::println, Throwable::printStackTrace, () -> System.out.println("Done"));

Slide 11

Slide 11 text

// expand to show full classes Observable.create(new OnSubscribe() { @Override public void call(Subscriber super String> subscriber) { subscriber.onNext("Hello World!"); subscriber.onCompleted(); } }).subscribe(new Subscriber() { @Override public void onCompleted() { System.out.println("Done"); } @Override public void onError(Throwable e) { e.printStackTrace(); } @Override public void onNext(String t) { System.out.println(t); } });

Slide 12

Slide 12 text

// add error propagation Observable.create(subscriber -> { try { subscriber.onNext("Hello World!"); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribe(System.out::println);

Slide 13

Slide 13 text

// add error propagation Observable.create(subscriber -> { try { subscriber.onNext(throwException()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribe(System.out::println);

Slide 14

Slide 14 text

// add error propagation Observable.create(subscriber -> { try { subscriber.onNext("Hello World!"); subscriber.onNext(throwException()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribe(System.out::println);

Slide 15

Slide 15 text

// add concurrency (manually) Observable.create(subscriber -> { new Thread(() -> { try { subscriber.onNext(getData()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).start(); }).subscribe(System.out::println); ?

Slide 16

Slide 16 text

// add concurrency (using a Scheduler) Observable.create(subscriber -> { try { subscriber.onNext(getData()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribeOn(Schedulers.io()) .subscribe(System.out::println); ? Learn more about parameterized concurrency and virtual time with Rx Schedulers at https://github.com/ReactiveX/RxJava/wiki/Scheduler

Slide 17

Slide 17 text

// add operator Observable.create(subscriber -> { try { subscriber.onNext(getData()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribeOn(Schedulers.io()) .map(data -> data + " --> at " + System.currentTimeMillis()) .subscribe(System.out::println); ?

Slide 18

Slide 18 text

// add error handling Observable.create(subscriber -> { try { subscriber.onNext(getData()); subscriber.onCompleted(); } catch (Exception e) { subscriber.onError(e); } }).subscribeOn(Schedulers.io()) .map(data -> data + " --> at " + System.currentTimeMillis()) .onErrorResumeNext(e -> Observable.just("Fallback Data")) .subscribe(System.out::println);

Slide 19

Slide 19 text

// infinite Observable.create(subscriber -> { int i=0; while(!subscriber.isUnsubscribed()) { subscriber.onNext(i++); } }).subscribe(System.out::println); Note: No backpressure support here. See Observable.from(Iterable) or Observable.range() for actual implementations

Slide 20

Slide 20 text

Hot Cold emits whether you’re ready or not examples mouse and keyboard events system events stock prices emits when requested (generally at controlled rate) examples database query web service request reading file Observable.create(subscriber -> { // register with data source }) Observable.create(subscriber -> { // fetch data })

Slide 21

Slide 21 text

Hot Cold emits whether you’re ready or not examples mouse and keyboard events system events stock prices emits when requested (generally at controlled rate) examples database query web service request reading file Observable.create(subscriber -> { // register with data source }) Observable.create(subscriber -> { // fetch data }) flow control flow control & backpressure

Slide 22

Slide 22 text

No content

Slide 23

Slide 23 text

No content

Slide 24

Slide 24 text

No content

Slide 25

Slide 25 text

Abstract Concurrency

Slide 26

Slide 26 text

Cold Finite Streams

Slide 27

Slide 27 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 28

Slide 28 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 29

Slide 29 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 30

Slide 30 text

 Observable  b  =  Observable.flatMap({  T  t  -­‐>            Observable  r  =  ...  transform  t  ...          return  r;    }) flatMap

Slide 31

Slide 31 text

 Observable  b  =  Observable.flatMap({  T  t  -­‐>            Observable  r  =  ...  transform  t  ...          return  r;    }) flatMap

Slide 32

Slide 32 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 33

Slide 33 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 34

Slide 34 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 35

Slide 35 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 36

Slide 36 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 37

Slide 37 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 38

Slide 38 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 39

Slide 39 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 40

Slide 40 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 41

Slide 41 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 42

Slide 42 text

       Observable.zip(a,  b,  (a,  b)  -­‐>  {                ...  operate  on  values  from  both  a  &  b  ...              return  Arrays.asList(a,  b);          }) zip { ( , ) }

Slide 43

Slide 43 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 44

Slide 44 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 45

Slide 45 text

No content

Slide 46

Slide 46 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 47

Slide 47 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 48

Slide 48 text

class  VideoService  {        def  VideoList  getPersonalizedListOfMovies(userId);        def  VideoBookmark  getBookmark(userId,  videoId);        def  VideoRating  getRating(userId,  videoId);        def  VideoMetadata  getMetadata(videoId);   } class  VideoService  {        def  Observable  getPersonalizedListOfMovies(userId);        def  Observable  getBookmark(userId,  videoId);        def  Observable  getRating(userId,  videoId);        def  Observable  getMetadata(videoId);   } ... create an observable api: instead of a blocking api ...

Slide 49

Slide 49 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); } 1 2 3 4 5 6

Slide 50

Slide 50 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); } 1 2 3 4 5 6

Slide 51

Slide 51 text

Non-Opinionated Concurrency

Slide 52

Slide 52 text

No content

Slide 53

Slide 53 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 54

Slide 54 text

No content

Slide 55

Slide 55 text

No content

Slide 56

Slide 56 text

No content

Slide 57

Slide 57 text

No content

Slide 58

Slide 58 text

public Observable handle(HttpServerRequest request, HttpServerResponse response) { // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }).flatMap(data -> { // output as SSE as we get back the data (no waiting until all is done) return response.writeAndFlush(new ServerSentEvent(SimpleJson.mapToJson(data))); }); }

Slide 59

Slide 59 text

No content

Slide 60

Slide 60 text

Decouples Consumption from Production // first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); })

Slide 61

Slide 61 text

// first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }) Decouples Consumption from Production

Slide 62

Slide 62 text

// first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }) Decouples Consumption from Production

Slide 63

Slide 63 text

Decouples Consumption from Production Event Loop API Request API Request API Request API Request API Request Dependency REST API new Command().observe() new Command(). observe() new Command(). observe() new Command(). observe() new Command(). observe() Collapser

Slide 64

Slide 64 text

// first request User object return getUser(request.getQueryParameters().get("userId")).flatMap(user -> { // then fetch personal catalog Observable> catalog = getPersonalizedCatalog(user) .flatMap(catalogList -> { return catalogList.videos().> flatMap(video -> { Observable bookmark = getBookmark(video); Observable rating = getRating(video); Observable metadata = getMetadata(video); return Observable.zip(bookmark, rating, metadata, (b, r, m) -> { return combineVideoData(video, b, r, m); }); }); }); // and fetch social data in parallel Observable> social = getSocialData(user).map(s -> { return s.getDataAsMap(); }); // merge the results return Observable.merge(catalog, social); }) ~5 network calls (#3 and #4 may result in more due to windowing) 1 2 3 4 5

Slide 65

Slide 65 text

Clear API Communicates Potential Cost class  VideoService  {        def  Observable  getPersonalizedListOfMovies(userId);        def  Observable  getBookmark(userId,  videoId);        def  Observable  getRating(userId,  videoId);        def  Observable  getMetadata(videoId);   }

Slide 66

Slide 66 text

class  VideoService  {        def  Observable  getPersonalizedListOfMovies(userId);        def  Observable  getBookmark(userId,  videoId);        def  Observable  getRating(userId,  videoId);        def  Observable  getMetadata(videoId);   } Implementation Can Differ BIO Network Call Local Cache Collapsed Network Call

Slide 67

Slide 67 text

class  VideoService  {        def  Observable  getPersonalizedListOfMovies(userId);        def  Observable  getBookmark(userId,  videoId);        def  Observable  getRating(userId,  videoId);        def  Observable  getMetadata(videoId);   } Implementation Can Differ and Change Local Cache Collapsed Network Call Collapsed Network Call BIO NIO Network Call

Slide 68

Slide 68 text

Retrieval, Transformation, Combination all done in same declarative manner

Slide 69

Slide 69 text

What about … ?

Slide 70

Slide 70 text

Error Handling

Slide 71

Slide 71 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).onErrorResumeNext(throwable -> { return Observable.just("fallback value"); }).subscribe(System.out::println);

Slide 72

Slide 72 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).onErrorReturn(throwable -> { return "fallback value"; }).subscribe(System.out::println);

Slide 73

Slide 73 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).retryWhen(attempts -> { return attempts.zipWith(Observable.range(1, 3), (throwable, i) -> i) .flatMap(i -> { System.out.println("delay retry by " + i + " second(s)"); return Observable.timer(i, TimeUnit.SECONDS); }).concatWith(Observable.error(new RuntimeException("Exceeded 3 retries"))); }) .subscribe(System.out::println, t -> t.printStackTrace());

Slide 74

Slide 74 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).retryWhen(attempts -> { return attempts.zipWith(Observable.range(1, 3), (throwable, i) -> i) .flatMap(i -> { System.out.println("delay retry by " + i + " second(s)"); return Observable.timer(i, TimeUnit.SECONDS); }).concatWith(Observable.error(new RuntimeException("Exceeded 3 retries"))); }) .subscribe(System.out::println, t -> t.printStackTrace());

Slide 75

Slide 75 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).retryWhen(attempts -> { return attempts.zipWith(Observable.range(1, 3), (throwable, i) -> i) .flatMap(i -> { System.out.println("delay retry by " + i + " second(s)"); return Observable.timer(i, TimeUnit.SECONDS); }).concatWith(Observable.error(new RuntimeException("Exceeded 3 retries"))); }) .subscribe(System.out::println, t -> t.printStackTrace());

Slide 76

Slide 76 text

Observable.create(subscriber -> { throw new RuntimeException("failed!"); }).retryWhen(attempts -> { return attempts.zipWith(Observable.range(1, 3), (throwable, i) -> i) .flatMap(i -> { System.out.println("delay retry by " + i + " second(s)"); return Observable.timer(i, TimeUnit.SECONDS); }).concatWith(Observable.error(new RuntimeException("Exceeded 3 retries"))); }) .subscribe(System.out::println, t -> t.printStackTrace());

Slide 77

Slide 77 text

Concurrency

Slide 78

Slide 78 text

Concurrency an Observable is sequential (no concurrent emissions) scheduling and combining Observables enables concurrency while retaining sequential emission

Slide 79

Slide 79 text

// merging async Observables allows each // to execute concurrently Observable.merge(getDataAsync(1), getDataAsync(2)) merge

Slide 80

Slide 80 text

// concurrently fetch data for 5 items Observable.range(0, 5).flatMap(i -> { return getDataAsync(i); })

Slide 81

Slide 81 text

Observable.range(0, 5000).window(500).flatMap(work -> { return work.observeOn(Schedulers.computation()) .map(item -> { // simulate computational work try { Thread.sleep(1); } catch (Exception e) {} return item + " processed " + Thread.currentThread(); }); })

Slide 82

Slide 82 text

Observable.range(0, 5000).buffer(500).flatMap(is -> { return Observable.from(is).subscribeOn(Schedulers.computation()) .map(item -> { // simulate computational work try { Thread.sleep(1); } catch (Exception e) {} return item + " processed " + Thread.currentThread(); }); })

Slide 83

Slide 83 text

Flow Control

Slide 84

Slide 84 text

Flow Control (backpressure)

Slide 85

Slide 85 text

no backpressure needed Observable.from(iterable).take(1000).map(i -> "value_" + i).subscribe(System.out::println);

Slide 86

Slide 86 text

Observable.from(iterable).take(1000).map(i -> "value_" + i).subscribe(System.out::println); no backpressure needed synchronous on same thread (no queueing)

Slide 87

Slide 87 text

Observable.from(iterable).take(1000).map(i -> "value_" + i) .observeOn(Schedulers.computation()).subscribe(System.out::println); backpressure needed

Slide 88

Slide 88 text

Observable.from(iterable).take(1000).map(i -> "value_" + i) .observeOn(Schedulers.computation()).subscribe(System.out::println); backpressure needed asynchronous (queueing)

Slide 89

Slide 89 text

Flow Control Options

Slide 90

Slide 90 text

Hot Cold emits whether you’re ready or not examples mouse and keyboard events system events stock prices emits when requested (generally at controlled rate) examples database query web service request reading file Observable.create(subscriber -> { // register with data source }) Observable.create(subscriber -> { // fetch data }) flow control flow control & backpressure

Slide 91

Slide 91 text

Block (callstack blocking and/or park the thread) Hot or Cold Streams

Slide 92

Slide 92 text

Temporal Operators (batch or drop data using time) Hot Streams

Slide 93

Slide 93 text

Observable.range(1, 1000000).sample(10, TimeUnit.MILLISECONDS).forEach(System.out::println); 110584 242165 544453 942880

Slide 94

Slide 94 text

Observable.range(1, 1000000).throttleFirst(10, TimeUnit.MILLISECONDS).forEach(System.out::println); 1 55463 163962 308545 457445 592638 751789 897159

Slide 95

Slide 95 text

Observable.range(1, 1000000).debounce(10, TimeUnit.MILLISECONDS).forEach(System.out::println); 1000000

Slide 96

Slide 96 text

Observable.range(1, 1000000).buffer(10, TimeUnit.MILLISECONDS) .toBlocking().forEach(list -> System.out.println("batch: " + list.size())); batch: 71141 batch: 49488 batch: 141147 batch: 141432 batch: 195920 batch: 240462 batch: 160410

Slide 97

Slide 97 text

No content

Slide 98

Slide 98 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] []

Slide 99

Slide 99 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] []

Slide 100

Slide 100 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] []

Slide 101

Slide 101 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] []

Slide 102

Slide 102 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] []

Slide 103

Slide 103 text

/* The following will emit a buffered list as it is debounced */ // first we multicast the stream ... using refCount so it handles the subscribe/unsubscribe Observable burstStream = intermittentBursts().take(20).publish().refCount(); // then we get the debounced version Observable debounced = burstStream.debounce(10, TimeUnit.MILLISECONDS); // then the buffered one that uses the debounced stream to demark window start/stop Observable> buffered = burstStream.buffer(debounced); // then we subscribe to the buffered stream so it does what we want buffered.toBlocking().forEach(System.out::println); [0, 1, 2] [0, 1, 2] [0, 1, 2, 3, 4, 5, 6] [0, 1, 2, 3, 4] [0, 1] [] https://gist.github.com/benjchristensen/e4524a308456f3c21c0b

Slide 104

Slide 104 text

Observable.range(1, 1000000).window(50, TimeUnit.MILLISECONDS) .flatMap(window -> window.count()) .toBlocking().forEach(count -> System.out.println("num items: " + count)); num items: 477769 num items: 155463 num items: 366768

Slide 105

Slide 105 text

Observable.range(1, 1000000).window(500000) .flatMap(window -> window.count()) .toBlocking().forEach(count -> System.out.println("num items: " + count)); num items: 500000 num items: 500000

Slide 106

Slide 106 text

Reactive Pull (dynamic push-pull)

Slide 107

Slide 107 text

Push (reactive) when consumer keeps up with producer. Switch to Pull (interactive) when consumer is slow. Bound all* queues.

Slide 108

Slide 108 text

*vertically, not horizontally Push (reactive) when consumer keeps up with producer. Switch to Pull (interactive) when consumer is slow. Bound all* queues.

Slide 109

Slide 109 text

Reactive Pull hot vs cold

Slide 110

Slide 110 text

Reactive Pull cold supports pull

Slide 111

Slide 111 text

Observable.from(iterable) Observable.from(0, 100000) Cold Streams emits when requested (generally at controlled rate) examples database query web service request reading file

Slide 112

Slide 112 text

Observable.from(iterable) Observable.from(0, 100000) Cold Streams emits when requested (generally at controlled rate) examples database query web service request reading file Pull

Slide 113

Slide 113 text

Reactive Pull hot receives signal

Slide 114

Slide 114 text

Reactive Pull hot receives signal *including Observables that don’t implement reactive pull support

Slide 115

Slide 115 text

hotSourceStream.onBackpressureBuffer().observeOn(aScheduler);

Slide 116

Slide 116 text

hotSourceStream.onBackpressureBuffer().observeOn(aScheduler);

Slide 117

Slide 117 text

hotSourceStream.onBackpressureBuffer().observeOn(aScheduler);

Slide 118

Slide 118 text

hotSourceStream.onBackpressureBuffer().observeOn(aScheduler);

Slide 119

Slide 119 text

hotSourceStream.onBackpressureDrop().observeOn(aScheduler);

Slide 120

Slide 120 text

stream.onBackpressure(strategy).subscribe

Slide 121

Slide 121 text

Hot Infinite Streams

Slide 122

Slide 122 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 123

Slide 123 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 124

Slide 124 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here) Hot Infinite Stream

Slide 125

Slide 125 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 126

Slide 126 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 127

Slide 127 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 128

Slide 128 text

(movieId) movieId=12345 movieId=34567

Slide 129

Slide 129 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 130

Slide 130 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 131

Slide 131 text

Stage 1 Stage 2 groupBy Event Streams sink

Slide 132

Slide 132 text

Stage 1 Stage 2 12345 56789 34567

Slide 133

Slide 133 text

Stage 1 Stage 2 12345 56789 34567

Slide 134

Slide 134 text

Stage 1 Stage 2 12345 56789 34567

Slide 135

Slide 135 text

Stage 1 Stage 2 12345 56789 34567

Slide 136

Slide 136 text

Stage 1 Stage 2 12345 56789 34567 12345 56789 34567

Slide 137

Slide 137 text

Stage 1 Stage 2 12345 56789 34567 12345 56789 34567

Slide 138

Slide 138 text

Stage 1 Stage 2 12345 56789 34567 12345 56789 34567

Slide 139

Slide 139 text

Stage 1 Stage 2 12345 56789 34567 12345 56789 34567

Slide 140

Slide 140 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 141

Slide 141 text

10mins || 1000 4:40-4:50pm 4:50-5:00pm 5:00-5:07pm (burst to 1000) 5:07-5:17pm

Slide 142

Slide 142 text

10mins || 1000 4:40-4:50pm 4:50-5:00pm 5:00-5:07pm (burst to 1000) 5:07-5:17pm

Slide 143

Slide 143 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 144

Slide 144 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 145

Slide 145 text

No content

Slide 146

Slide 146 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 147

Slide 147 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 148

Slide 148 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 149

Slide 149 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 150

Slide 150 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 151

Slide 151 text

MantisJob .source(NetflixSources.moviePlayAttempts()) .stage(playAttempts -> { return playAttempts.groupBy(playAttempt -> { return playAttempt.getMovieId(); }) }) .stage(playAttemptsByMovieId -> { playAttemptsByMovieId .window(10,TimeUnit.MINUTES, 1000) // buffer for 10 minutes, or 1000 play attempts .flatMap(windowOfPlayAttempts -> { return windowOfPlayAttempts .reduce(new FailRatioExperiment(playAttemptsByMovieId.getKey()), (experiment, playAttempt) -> { experiment.updateFailRatio(playAttempt); experiment.updateExamples(playAttempt); return experiment; }).doOnNext(experiment -> { logToHistorical("Play attempt experiment", experiment.getId(),experiment); // log for offline analysis }).filter(experiment -> { return experiment.failRatio() >= DYNAMIC_PROP("fail_threshold").get(); }).map(experiment -> { return new FailReport(experiment, runCorrelations(experiment.getExamples())); }).doOnNext(report -> { logToHistorical("Failure report", report.getId(), report); // log for offline analysis }) }) }) .sink(Sinks.emailAlert(report -> { return toEmail(report)})) // anomalies trigger events (simple email here)

Slide 152

Slide 152 text

stream.onBackpressure(strategy?).subscribe

Slide 153

Slide 153 text

stream.onBackpressure(buffer).subscribe

Slide 154

Slide 154 text

stream.onBackpressure(drop).subscribe

Slide 155

Slide 155 text

stream.onBackpressure(sample).subscribe

Slide 156

Slide 156 text

stream.onBackpressure(scaleHorizontally).subscribe

Slide 157

Slide 157 text

Reactive-Streams https://github.com/reactive-streams/reactive-streams

Slide 158

Slide 158 text

Reactive-Streams https://github.com/reactive-streams/reactive-streams https://github.com/ReactiveX/RxJavaReactiveStreams

Slide 159

Slide 159 text

final ActorSystem system = ActorSystem.create("InteropTest"); final FlowMaterializer mat = FlowMaterializer.create(system); // RxJava Observable Observable> oddAndEvenGroups = Observable.range(1, 1000000) .groupBy(i -> i % 2 == 0) .take(2); Observable strings = oddAndEvenGroups. flatMap(group -> { // schedule odd and even on different event loops Observable asyncGroup = group.observeOn(Schedulers.computation()); // convert to Reactive Streams Publisher Publisher groupPublisher = RxReactiveStreams.toPublisher(asyncGroup); // convert to Akka Streams Source and transform using Akka Streams ‘map’ and ‘take’ operators Source stringSource = Source.from(groupPublisher).map(i -> i + " " + group.getKey()).take(2000); // convert back from Akka to Rx Observable return RxReactiveStreams.toObservable(stringSource.runWith(Sink. fanoutPublisher(1, 1), mat)); }); strings.toBlocking().forEach(System.out::println); system.shutdown(); compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'com.typesafe.akka:akka-stream-experimental_2.11:0.10-M1'

Slide 160

Slide 160 text

final ActorSystem system = ActorSystem.create("InteropTest"); final FlowMaterializer mat = FlowMaterializer.create(system); // RxJava Observable Observable> oddAndEvenGroups = Observable.range(1, 1000000) .groupBy(i -> i % 2 == 0) .take(2); Observable strings = oddAndEvenGroups. flatMap(group -> { // schedule odd and even on different event loops Observable asyncGroup = group.observeOn(Schedulers.computation()); // convert to Reactive Streams Publisher Publisher groupPublisher = RxReactiveStreams.toPublisher(asyncGroup); // convert to Akka Streams Source and transform using Akka Streams ‘map’ and ‘take’ operators Source stringSource = Source.from(groupPublisher).map(i -> i + " " + group.getKey()).take(2000); // convert back from Akka to Rx Observable return RxReactiveStreams.toObservable(stringSource.runWith(Sink. fanoutPublisher(1, 1), mat)); }); strings.toBlocking().forEach(System.out::println); system.shutdown(); compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'com.typesafe.akka:akka-stream-experimental_2.11:0.10-M1'

Slide 161

Slide 161 text

// RxJava Observable Observable> oddAndEvenGroups = Observable.range(1, 1000000) .groupBy(i -> i % 2 == 0) .take(2); Observable strings = oddAndEvenGroups. flatMap(group -> { // schedule odd and even on different event loops Observable asyncGroup = group.observeOn(Schedulers.computation()); // convert to Reactive Streams Publisher Publisher groupPublisher = RxReactiveStreams.toPublisher(asyncGroup); // Convert to Reactor Stream and transform using Reactor Stream ‘map’ and ‘take’ operators Stream linesStream = Streams.create(groupPublisher).map(i -> i + " " + group.getKey()).take(2000); // convert back from Reactor Stream to Rx Observable return RxReactiveStreams.toObservable(linesStream); }); strings.toBlocking().forEach(System.out::println); compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'org.projectreactor:reactor-core:2.0.0.M1'

Slide 162

Slide 162 text

// RxJava Observable Observable> oddAndEvenGroups = Observable.range(1, 1000000) .groupBy(i -> i % 2 == 0) .take(2); Observable strings = oddAndEvenGroups. flatMap(group -> { // schedule odd and even on different event loops Observable asyncGroup = group.observeOn(Schedulers.computation()); // convert to Reactive Streams Publisher Publisher groupPublisher = RxReactiveStreams.toPublisher(asyncGroup); // Convert to Reactor Stream and transform using Reactor Stream ‘map’ and ‘take’ operators Stream linesStream = Streams.create(groupPublisher).map(i -> i + " " + group.getKey()).take(2000); // convert back from Reactor Stream to Rx Observable return RxReactiveStreams.toObservable(linesStream); }); strings.toBlocking().forEach(System.out::println); compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'org.projectreactor:reactor-core:2.0.0.M1'

Slide 163

Slide 163 text

compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'io.ratpack:ratpack-rx:0.9.10' try { RatpackServer server = EmbeddedApp.fromHandler(ctx -> { Observable o1 = Observable.range(0, 2000) .observeOn(Schedulers.computation()).map(i -> { return "A " + i; }); Observable o2 = Observable.range(0, 2000) .observeOn(Schedulers.computation()).map(i -> { return "B " + i; }); Observable o = Observable.merge(o1, o2); ctx.render( ServerSentEvents.serverSentEvents(RxReactiveStreams.toPublisher(o), e -> e.event("counter").data("event " + e.getItem())) ); }).getServer(); server.start(); System.out.println("Port: " + server.getBindPort()); } catch (Exception e) { e.printStackTrace(); }

Slide 164

Slide 164 text

compile 'io.reactivex:rxjava:1.0.+' compile 'io.reactivex:rxjava-reactive-streams:0.3.0' compile 'io.ratpack:ratpack-rx:0.9.10' try { RatpackServer server = EmbeddedApp.fromHandler(ctx -> { Observable o1 = Observable.range(0, 2000) .observeOn(Schedulers.computation()).map(i -> { return "A " + i; }); Observable o2 = Observable.range(0, 2000) .observeOn(Schedulers.computation()).map(i -> { return "B " + i; }); Observable o = Observable.merge(o1, o2); ctx.render( ServerSentEvents.serverSentEvents(RxReactiveStreams.toPublisher(o), e -> e.event("counter").data("event " + e.getItem())) ); }).getServer(); server.start(); System.out.println("Port: " + server.getBindPort()); } catch (Exception e) { e.printStackTrace(); }

Slide 165

Slide 165 text

RxJava 1.0 Final November 18th

Slide 166

Slide 166 text

Mental Shift imperative → functional sync → async pull → push

Slide 167

Slide 167 text

Concurrency and async are non-trivial. Rx doesn’t trivialize it. Rx is powerful and rewards those who go through the learning curve.

Slide 168

Slide 168 text

Single Multiple Sync T getData() Iterable getData() Stream getData() Async Future getData() Observable getData()

Slide 169

Slide 169 text

Abstract Concurrency

Slide 170

Slide 170 text

Non-Opinionated Concurrency

Slide 171

Slide 171 text

Decouple Production from Consumption

Slide 172

Slide 172 text

Powerful Composition of Nested, Conditional Flows

Slide 173

Slide 173 text

First-class Support of Error Handling, Scheduling & Flow Control

Slide 174

Slide 174 text

RxJava http://github.com/ReactiveX/RxJava http://reactivex.io

Slide 175

Slide 175 text

Reactive Programming in the Netflix API with RxJava http://techblog.netflix.com/2013/02/rxjava-netflix-api.html Optimizing the Netflix API http://techblog.netflix.com/2013/01/optimizing-netflix-api.html Reactive Extensions (Rx) http://www.reactivex.io Reactive Streams https://github.com/reactive-streams/reactive-streams Ben Christensen @benjchristensen RxJava https://github.com/ReactiveX/RxJava @RxJava RxJS http://reactive-extensions.github.io/RxJS/ @ReactiveX jobs.netflix.com