When building microservices or web apps, we often take the path of least resistance and go with a stateless approach. The justification is that going the stateful route is too hard and too complicated. Based on the state of the tools that we typically use to build apps, going stateless is a wise decision given that the commonly used backend toolsets and frameworks tend to shy away from dealing with distributed, clustered systems. However, with the spectacular rise of Kubernetes, many developers are diving head first into the clustered world. This mass migration to the clustered, scalable, and resilient Kubernetes playing field opens up new opportunities for how we build systems. One of the new ways of doing things is the actor model. In the pre-Kubernetes world, everything is an object; in the post-Kubernetes world, everything is an actor. Actors are fundamental building blocks, like objects, that are stateful, are inherently concurrent, and with the Akka Toolkit, systems of actors naturally exist and collaborate in clustered environments. In this talk, we will explore some theory and code of a live actor system based microservice running in a clustered Kubernetes environment.

1. Hugh McKee (@mckeeh3), Developer Advocate
Building Stateful Clustered Microservices
with Java, Actors, and Kubernetes

2. Hugh McKee (@mckeeh3), Developer Advocate

3. Hugh McKee (@mckeeh3), Developer Advocate

4. Hugh McKee (@mckeeh3), Developer Advocate

5. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

6. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

7. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

8. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

9. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

10. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

11. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

12. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

13. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

14. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

15. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

16. The actor model in computer science is a mathematical model of
concurrent computation that treats "actors" as the universal primitives
of concurrent computation. In response to a message that it receives,
an actor can: make local decisions, create more actors, send more
messages, and determine how to respond to the next message
received. Actors may modify their own private state, but can only affect
each other through messages (avoiding the need for any locks).
Wikipedia

17. Kubernetes (K8s) is an open-source system for
automating deployment, scaling, and management
of containerized applications.

18. Kubernetes (K8s) is an open-source system for
automating deployment, scaling, and management
of containerized applications.
elastic
resilient

19. Akka cluster and Kubernetes
Live demo

20. “Crop Circle”
Shows
Running Pods / JVMs

21. “Crop Circle”
Shows
Running Pods / JVMs
POD / JVM
POD / JVM
POD / JVM

22. Entity
Shard
Singleton
HTTP Server
Pod

23. Entities added as needed
Inactive entities shutdown

24. Microservice Application
Microservice

25. Resilience and Scale Demo

26. The enabler of these characteristics is a Cloud-Ready Message Driven Model.
Lightbend codified these principles into the Reactive Manifesto in 2013; 24,000+ signatories around the world so far.
Responsive Resilient Elastic
React to
Users
React to
Failures
React to
Load Variance
Low latency / High
performance
Real-time / NRT
Graceful, Non-catastrophic
Recovery
Self-Healing
Responsive in the face of
changing loads
Reactive Systems

27. The enabler of these characteristics is a Cloud-Ready Message Driven Model.
Lightbend codified these principles into the Reactive Manifesto in 2013; 24,000+ signatories around the world so far.
Responsive Resilient Elastic
React to
Users
React to
Failures
React to
Load Variance
Low latency / High
performance
Real-time / NRT
Graceful, Non-catastrophic
Recovery
Self-Healing
Responsive in the face of
changing loads
Reactive Systems

28. Java Maven K8
Project Review

29. Clustered Entity Actors

30. You
entity 64
Load Balancer
Me
entity 17
Other
entity 76

31. You
entity 64
Load Balancer
Me
entity 17
Other
entity 76

32. Clustered Entity Actors
Code Review

33. Load Balancer
Pod 1

34. Load Balancer
Pod 2Pod 1 Pod 3

35. Load Balancer
Pod 2Pod 1 Pod 3

37. Distributed Actors
Cluster Aware
Actors

38. Distributed Actors
Cluster Aware
Actors

39. Distributed Actors
Cluster Aware
Actors

40. Distributed Actors
Cluster Aware
Actors

41. Distributed Actors
Cluster Aware
Actors

42. Distributed Actors
Cluster Aware
Actors

43. Cluster Aware Actors
Code Review

44. Shard Distribution
Akka Cluster Singleton

45. Shard Distribution
Akka Cluster Singleton

46. Shard Distribution
Akka Cluster Singleton

47. Entity Distributed Sharding
Akka Cluster Sharding

48. Entity Distributed
Sharding
Akka Cluster Sharding

49. Event Sourcing & CQRS
Akka Persistence & Persistence Query

50. Event Sourcing & CQRS
Akka Persistence &
Persistence Query

51. User ID Time Event
You 64 08:11 Add Item 1567
Me 17 08:15 Add Item 3254
Other 76 08:16 Add Item 8359
You 64 08:20 Add Item 2632
Me 17 08:20 Add Item 4983
Other 76 08:24 Change Item 8359
You 64 08:25 Remove Item 1567
You 64 08:26 Add shipping address
Other 76 08:30 Add Item 2438
Me 17 08:33 Add shipping address
Me 17 08:33 Add billing address
You 64 08:35 Add billing address
Event Sourcing & CQRS

52. You
entity 64
Load Balancer
Me
entity 17
Other
entity 76

53. Akka Reactive Systems and Kubernetes
Responsive Resilient Elastic
React to
Users
React to
Failures
React to
Load Variance
Low latency / High
performance
Real-time / NRT
Graceful, Non-catastrophic
Recovery
Self-Healing
Responsive in the face of
changing loads
a beautiful relationship

54. /akka-java-cluster-kubernetes
/akka-java-cluster.git
/akka-java-cluster-aware.git
/akka-java-cluster-singleton.git
/akka-java-cluster-sharding.git
/akka-java-cluster-persistence.git
/akka-java-cluster-persistence-query.git
https://github.com/mckeeh3…

55. Upgrade your grey matter!
Get the free O’Reilly book by Hugh McKee,
Developer Advocate at Lightbend
https://www.lightbend.com/ebooks

56. Hugh McKee (@mckeeh3), Developer Advocate
hugh.mckee@lightbend.com
Akka and Kubernetes:
The beginning of a beautiful relationship