Akka in action - W-JAX 2012

Transcript

1. Akka in action W-JAX 2012 Heiko Seeberger
2. What is Akka?
3. What is Akka? • Akka is a mountain in northern Sweden
4. What is Akka? • Akka is a mountain in northern Sweden • Akka is part of the Typesafe Stack
5. What is Akka? • Akka is a mountain in northern Sweden • Akka is part of the Typesafe Stack • Akka is a toolkit and runtime for scalable systems
6. What is scalability?
7. What is scalability? • Ability of a system to handle growing work or
8. What is scalability? • Ability of a system to handle growing work or • its ability to be enlarged to accommodate growth.
9. Why should we care about scalability?
10. Because of the human nature
11. How can we build scalable systems?
12. Scaling up
13. Scaling up Concurrency
14. Scaling out
15. Scaling out Distribution
16. How can we easily build scalable systems?
18. Akka’s vision
19. Akka’s vision • Simpler concurrency (scale up)
20. Akka’s vision • Simpler concurrency (scale up) • Simpler distribution (scale out)
21. Akka’s vision • Simpler concurrency (scale up) • Simpler distribution (scale out) • Simpler fault-tolerance (self healing)
22. Akka’s vision • Simpler concurrency (scale up) • Simpler distribution (scale out) • Simpler fault-tolerance (self healing) • All of that with a single unified programming model
23. The Akka toolkit
24. The Akka toolkit • Akka runs on the JVM
25. The Akka toolkit • Akka runs on the JVM • Akka can be used from Java and Scala
26. The Akka toolkit • Akka runs on the JVM • Akka can be used from Java and Scala • Akka can be integrated with common infrastructure, e.g. Spring, Camel, ZeroMQ, etc.
27. What’s inside the toolbox?
28. What’s inside the toolbox? • Actors
29. What’s inside the toolbox? • Actors • Agents
30. What’s inside the toolbox? • Actors • Agents • Dataflow concurrency
31. What’s inside the toolbox? • Actors • Agents • Dataflow concurrency • Software Transactional Memory (STM)
32. What is an actor?
33. What is an actor? • Carl Hewitt (1973): Fundamental unit of computation
34. What is an actor? • Carl Hewitt (1973): Fundamental unit of computation • Behavior - react on messages it receives
35. What is an actor? • Carl Hewitt (1973): Fundamental unit of computation • Behavior - react on messages it receives • State - shielded from the rest of the world, no need for synchronization
36. What is an actor? • Carl Hewitt (1973): Fundamental unit of computation • Behavior - react on messages it receives • State - shielded from the rest of the world, no need for synchronization • Communication - interact with other actors exclusively via messages
37. What is an actor? • Carl Hewitt (1973): Fundamental unit of computation • Behavior - react on messages it receives • State - shielded from the rest of the world, no need for synchronization • Communication - interact with other actors exclusively via messages • “One actor is no actor” - they come in systems
38. Other notions of an actor
39. Other notions of an actor • Implementation of message passing concurrency
40. Other notions of an actor • Implementation of message passing concurrency • Isolated lightweight processes
41. Other notions of an actor • Implementation of message passing concurrency • Isolated lightweight processes • Asynchronous and non-blocking
42. The Akka Actor
43. The Akka Actor
44. The Akka Actor
45. The Akka Actor
46. Receive messages
47. Receive messages public class Hello extends UntypedActor { @Override public void onReceive(Object message) { System.out.println("Hello, world!"); } }
48. Receive messages public class Hello extends UntypedActor { public int count = 0; @Override public void onReceive(Object message) { count += 1; System.out.println( "Hello for the " + count + " time!" ); } }
49. Receive messages public class Hello extends UntypedActor { public int count = 0; @Override public void onReceive(Object message) { count += 1; System.out.println( "Hello for the " + count + " time!" ); } } One at a time
50. Receive messages public class Hello extends UntypedActor { public int count = 0; @Override public void onReceive(Object message) { count += 1; System.out.println( "Hello for the " + count + " time!" ); } } One at a time Thread-safe
51. Send messages
52. Send messages ActorRef hello = ... hello.tell("Hi!");
53. Send messages ActorRef hello = ... hello.tell("Hi!"); Asynchronous and nonblocking
54. The Akka ActorSystem
55. The Akka ActorSystem
56. The Akka ActorSystem
57. Create an actor system
58. Create an actor system ActorSystem system = ActorSystem.create("hello");
59. Create a top-level actor
60. Create a top-level actor ActorRef hello = system.actorOf(new Props(new Hello()));
61. Create a child actor
62. Create a child actor ActorRef child = context().actorOf(new Props(new Child()));
63. Embrace failure
64. Embrace failure • Let it crash!
65. Embrace failure • Let it crash! • Supervision: Like in real life, parents supervise their children
66. Embrace failure • Let it crash! • Supervision: Like in real life, parents supervise their children • Depending on the parent’s supervisor strategy, failing actors can get stopped, restarted or resumed
67. HakkyHour demo
68. The HakkyHour bar
69. The HakkyHour bar • Our drinks: Akkarita, MaiPlay, PinaScalada
70. The HakkyHour bar • Our drinks: Akkarita, MaiPlay, PinaScalada • Our actors: guests, waiters, head waiter, barkeepers
71. The HakkyHour bar • Our drinks: Akkarita, MaiPlay, PinaScalada • Our actors: guests, waiters, head waiter, barkeepers • Our messages: Order, drink served, complaint, etc.
72. The HakkyHour bar • Our drinks: Akkarita, MaiPlay, PinaScalada • Our actors: guests, waiters, head waiter, barkeepers • Our messages: Order, drink served, complaint, etc. • Our failures: Guest drunk, waiter frustrated
73. HakkyHour actor system
74. HakkyHour messages
75. HakkyHour messages
76. HakkyHour messages
77. HakkyHour messages
78. HakkyHour messages
79. HakkyHour messages
80. HakkyHour messages
81. Live demo ...
82. Why does Akka make building scalable systems easy?
83. Benefits of using Akka actors
84. Benefits of using Akka actors • You don’t have to deal with low-level concurrency details
85. Benefits of using Akka actors • You don’t have to deal with low-level concurrency details • You can manage failures easily
86. Benefits of using Akka actors • You don’t have to deal with low-level concurrency details • You can manage failures easily • Distribution is just a deployment decision (not covered here)
87. Questions?
88. Thank you! Code: github.com/hseeberger/hakkyhour Twitter: @hseeberger