Full-Stack, Message-oriented Programming w/ Akka.NET Actors
7 likes1,957 views
This document discusses Akka.NET, a framework for building distributed, concurrent applications using the actor model. It explains that in the actor model, everything is an actor that can send and receive immutable messages asynchronously. Actors in Akka.NET are lightweight threads that handle messages concurrently without shared state. The document provides examples of how to define actors and send messages in Akka.NET. It argues that the actor model enables building highly concurrent, scalable and fault-tolerant applications more easily than traditional threading approaches.
Full-Stack, Message-oriented Programming w/ Akka.NET Actors
- 1. Full-Stack, Message-Oriented Programming With Akka.NET Actors Andrew Skotzko Co-founder, Petabridge Contributor, Akka.NET @askotzko
- 2. My Wish For You
- 3. Messaging
- 5. Concurrency. Why should we care?
- 6. 1973–2005
- 11. What the hell are you talking about?
- 12. Concurrency ...time to embrace it.
- 13. Why is this hard?
- 14. Writing Multithreaded Code → Make classes/functions → Find places we can do multiple things at once → Queue work onto new thread / threadpool → Use synchronization mechanisms (protect shared mem) → Finish work on one thread; consume on another
- 15. Shared State Programming Sucks → Synchronizing shared state = pit of doom → Coordinating work btw. threads → Debugging multi-threaded code...yay! → Third-party libraries → ...Writing great thread-safe code is hard.
- 16. Random? 1. .NET 4.0: TPL (2010) 2. .NET 4.5: async/await (2012)
- 17. There. must. be. a. better. way.
- 19. What is the actor model?
- 20. 1973
- 22. Everything is an actor.
- 23. Okay.. so what is an actor?
- 24. 3 Core Abilities 1. Send messages 2. Create other actors 3. Change behavior
- 25. Messages
- 27. Sending a Message myActorRef.Tell("this is my message");
- 28. Messages → Must be immutable → Processed serially
- 31. Change Behavior
- 35. Actors Live at Unique Addresses
- 38. Which means...
- 47. This Is An Akka.NET Actor public class FooActor : UntypedActor { public FooActor() { // message handling code goes here... } }
- 48. These Are Messages // just POCOs public class Hello {} public class NewOrder { public NewOrder (string name) { Username = name; UserId = Guid.NewGuid (); } public Guid UserId { get; private set;} public string Username {get; private set;} }
- 51. Where can I use this?
- 52. → Analytics systems → Marketing automation → Multi-player games → Device / IoT tracking → Alerting & monitoring systems → Recommendation engines → Dynamic pricing → Scalable microservices → Reactive UI
- 53. Wait... How Can This Be Massively Concurrent?
- 58. Message Buses & Actors
- 60. How Are They Different?
- 63. How Do They Go Together?
- 64. Akka.NET is a great consumer/producer for NServiceBus
- 66. async by default
- 67. p2p mesh
- 68. Queued Messages
- 69. Extensible
- 72. Pub-Sub
- 74. public class PubSubActor : ReceiveActor { protected HashSet<IActorRef> Subscribers = new HashSet<IActorRef>(); public PubSubActor() { // add new subscriber Receive<Subscribe>(subscribe => Subscribers.Add(subscribe.Subscriber)); // remove subscriber Receive<Unsubscribe>(unsubscribe => Subscribers.Remove(unsubscribe.Subscriber)); // publish status to subscribers Receive<JobStats>(stats => { TotalStats = TotalStats.Merge(stats); PublishStatus(); }); } // push latest status into subscribers private void PublishStatus() { foreach (var sub in Subscribers) sub.Tell(RunningStatus); } }
- 78. Intuitive code
- 80. Efficient Use of Hardware
- 81. One Case Study → 85% reduction in code base → 60% reduction in hardware needed → "Code just lets us focus on the problem"
- 83. Location transparency + Durability + Easy, thread-safe scale out