From Web to Flux @DevoxxBE 2023.pptx
Download as PPTX, PDF0 likes286 views
Ready for a deep dive into the world's most challenging programming paradigm? Reactive programming can simplify asynchronous and event-driven applications, but without a strong understanding, it can lead to frustration, recurring patchwork, missed deadlines, and costly bugs. In this intensive three-hour session, we'll transition a traditional Spring application to WebFlux, revealing patterns and aanti-patterns when working with repositories, REST APIs, queues, and legacy libraries. You'll gain a clear understanding of often overlooked but critical aspects like subscribe signal, errors, cancellation, and signal loss. As a bonus, we'll debate the future of Reactive vs Virtual Threads, production-ready in Java 21. This session is crucial for developers already working with reactive programming or those intending to make the leap.
Ad
More Related Content
What's hot (20)
Similar to From Web to Flux @DevoxxBE 2023.pptx (20)
Ad
More from Victor Rentea (20)
Ad
Recently uploaded (20)
![Get & Download Wondershare Filmora Crack Latest [2025]](https://cdn.slidesharecdn.com/ss_thumbnails/revolutionizingresidentialwi-fi-250422112639-60fb726f-250429170801-59e1b240-thumbnail.jpg?width=560&fit=bounds)
From Web to Flux @DevoxxBE 2023.pptx
- 2. victorrentea.ro/training-offer 👋 Hi, I'm Victor Rentea 🇷🇴 PhD(CS): VictorRentea.ro Java Champion, 18 years of code, 10 years of teaching Consultant & Trainer for 120+ companies: ❤️ Clean Code, Architecture, Unit Testing 🛠 Spring, Hibernate, Reactive/WebFlux ⚡️ Java Performance, Secure Coding 🔐 Lots of Conference Talks on YouTube Founder of European Software Crafters Community (6K members) 🔥 Free 1-hour webinars, after work 👉 victorrentea.ro/community Past events on youtube.com/vrentea Father of 👧👦, servant of a 🐈, weekend gardener 🌼 VictorRentea.ro
- 3. 3 Reactive Programming unbounded if no backpressure
- 4. 4 tl;dr This session is not for children !
- 5. 5 Crash Course in Reactive Programming
- 6. 6 Reactive Programming is programming with asynchronous data streams Watch: The introduction to Reactive Programming you've been missing - https://gist.github.com/staltz/868e7e9bc2a7b8c1f754
- 7. 7
- 8. 8
- 9. 9
- 10. 10
- 11. 11
- 12. 12
- 13. 13
- 15. 15 Operator
- 16. 16 Subscriber upstream ↑ Publisher downstream ↓ Operator
- 17. 17 .buffer(Duration.ofSeconds(2)) .flatMap(postIds -> postRepo.findAllById(postIds)) .map(LikedPosts::new) Subscriber upstream ↑ Publisher downstream ↓ Operator
- 18. 18 Project Reactor is for Backend Java what RxJava is for Android
- 19. 20 If you have a Spring MVC application that works fine, there is no need to change. Imperative programming is the easiest way to write, understand, and debug code. You have maximum choice of libraries, since most are blocking. If you have a large team, keep in mind the steep learning curve in the shift to non-blocking, functional, and declarative programming. A practical way to start without a full switch is to use the reactive WebClient. Beyond that, start small and measure the benefits. We expect that, for a wide range of applications, the shift is unnecessary. https://docs.spring.io/spring-framework/docs/current/reference/html/web-reactive.html#webflux-framework-choice
- 20. 21
- 21. 22 - the whole story -
- 22. 23
- 23. 24 Flux Emits 0..N items🔵 + COMPLETE| or ERROR❌ ... Many elements Flux.just(1,2) Flux.fromIterable(list) One element Flux.just("a") No element Flux.empty() Infinite Flux ... Flux.interval() KafkaReceiver.receive() ... Error Flux.error() Δt time
- 24. 25
- 25. 26 Mono Emits 0..1 items🔵 + COMPLETE| or ERROR❌ No element Mono.empty() One item Mono.just(1) error Mono.error(new Ex(..)) Best Practice: If a function return nothing declare its return type: Mono<Void> eg. sendEmail(..): Mono<Void> Data items cannot be null in Reactor; instead, return a Mono.empty(). ⚠️ How do operator handle empty?
- 26. 27 Marble Diagrams (THE tool to self-documenting about operators)
- 28. 29 Fundamental Reactive Programming Rules Nothing happens until you subscribe Avoid .subscribe()/.block() Instead, return a Publisher (Mono/Flux)
- 29. 30 CodE
- 30. 31
- 31. 32 © VictorRentea.ro a training by Spring WebFlux Lessons • A method calling network must return a Publisher. Network call is sent at subscribe↑ • A method returning Publisher must not BLOCK nor THROW (return an error() instead) • BLOCK is only legal via fromCallable({blocking code}).subscribedOn(boundedElastic()) • Every cold Publisher must be subscribed (once!) for a network call to happen • Avoid keeping Publisher instances in variables ⤴️ • Mind the signals in Marble Diagrams⚠️: behavior on empty↓, cancel↑, subscribe↑ • Request metadata 🪄🎩 is propagated via ReactorContext instead of ThreadLocal • Return your Publishers to Spring: to subscribe, pass RC⤴️ and manage cancellation • Avoid doing .subscribe() and .block() ⤴️, except at startup, blocking MQ listeners, timers • Master Functional Programming 🪄, Favor immutability to avoid race bugs💀 • Learn Reactor⚛️ operators, avoid excessive nested lambdas , small functions❤️ • Take it baby-steps: start migration with WebClient..block() and Mono.fromRunnable() • Infinite Flux/Sink: beware of cancellation⚠️
- 32. 33 Default way of handling HTTP requests Boundary System Service A Service B DB time API call (http request) thread released (http response sent back) thread blocked ( 1 / 200 in the thread pool ) WASTE OF RESOURCES STARVATION RISK API call (http request) API call (http request) Query (JDBC call) ⭐️ Fixed by Java 21
- 33. 34 © VictorRentea.ro a training by Virtual Threads (Project Loom) (included in Java 19 in preview) When a Virtual Thread blocks for any reason (eg. DB, API call, File access, Locks), the JVM reuses the underlying OS thread to execute other runnable virtual threads Therefore, a Virtual Thread is very cheap Virtual Threads are very light: • We can take 100.000s requests in parallel even if our code blocks • No need for thread pools anymore • Structured concurrency can be used to fork/join threads to handle cancellations • ThreadLocal data needs to be kept small • Can pin on an OS thread or Can be unfair if non-interruptible CPU work
- 34. 35 © VictorRentea.ro a training by Virtual Threads can handle huge number of independent parallel requests, without the risks of Reactive Programming ☠️ Use Cases for Reactive Programming: Handling async streams of data signals: IoT, MQ Server <-> Browser bidirectional comm Streaming large data in transit
Editor's Notes
- #3: I'm victor rentea from Romania. I'm a java champion, working in our field for 17 years. 8 years ago I realized coding was not enough for me, and I started looking around to help the others. Today this has become my full-time job: training and consultancy for companies throughout Europe. My favorite topics are ... but of course, to talk about these topics you have to master the frameworks you use, so I do intense workshops on Spring Framework, .... More senior groups often call me for performance tuning or secure coding. If you want to know more, you can find there my full training offer Besides the talks at different conferences that you can find online, I try to to one webinar each month for my community. A few years ago I started this group on meetup to have where to share my ideas and learn from the others in turn. - This community has exceeded my wildest dreams, turning into one of the largest communities in the world on Software Craftsmanship. - So what happens there? One day a month we have a zoom online webinar of 1-2 hours after work, during which we discuss one topic and then debate various questions from the participants – usually we have close to 100 live participants, so it's very engaging. If you want to be part of the fun, DO join us, it's completely free. - Many past events are available on my youtube channel. - Outside of work, I have 2 kids and a cat that wakes us up in the middle of the night.