A problem-solution approach to delivering Microservice-based design using .NET and RabbitMQ, based on real-world examples. The focus highlights key design decisions, custom problems encountered, lessons learned, and general advice. This is technical presentation, for the most part, also briefly touching on softer elements.

1. Microservice Architecture
Beginning Microservices with .NET & RabbitMQ
Paul Mooney
@daishisystems | www.insidethecpu.com

2. Overview
 Microservice design based on real-world examples
 Focus on, but not restricted to .NET and RabbitMQ
 Design Patterns
 Decoupled Middleware
 Circuit Breaker
 Flyball Governor
 Implementation Problems
 Consistent Message Delivery
 Load balancing a RabbitMQ Cluster
 Selling to the Business

3. Decoupled MiddlewareDispatcherReceiver
ASP.NET
ReceiverDispatcher
SimpleMathMicroservice
Input Queue
Output Queue
Message Bus

4. Decoupled Middleware
Microservice Daemon
Message Dispatcher
Event Listener
Microservice Daemon
Event Listener
Message Dispatcher
Service Bus
Queue #1
Queue #2
Message Message
Message Message

5. Circuit Breaker
Filter
CircuitBreaker
Service
Feature1Feature2Feature3
Client
Request

6. Circuit Breaker
Filter
CircuitBreaker
Service
Feature1Feature2Feature3
Client
requestCount < threshHold

7. Circuit Breaker
Filter
CircuitBreaker
Service
Feature1Feature2Feature3
Client
Request
requestCount >= threshHold

8. Flyball Governor

9. Flyball Governor
 Load never drops to 0% capacity
 Load never rises to 100% capacity
 Balance is achieved through polling
 Requires detailed tuning to establish optimal setting
 End result is Autoscaling through natural equilibrium

10. Flyball Governor
QueueWatch
Inbound Queue
Outbound Queue
Diagnostics Queue
Inbound Queue
Outbound Queue
Diagnostics Queue
Inbound Queue
Outbound Queue
Diagnostics Queue
1
2 2 2
Direct Exchange
3 3 3

11. Flyball GovernorDispatcherReceiver
ASP.NET
ReceiverDispatcher
SimpleMathMicroservice
Input Queue
Output Queue
Message Bus

12. Flyball GovernorDispatcherReceiver
ASP.NET
Input Queue
Output Queue
Message Bus
ReceiversDispatchers
SimpleMathMicroservices 1...n

13. Microservices as N-Tier Applications
AMQP
AMQP
AMQP
AMQP
Browser API Queue Microservice Instances
HttpRequestMapper
CommandMessageBuilder
MessageDispatcher
ResponseMessageParser
ResponseMessageConsumer CommandMessageParser
CommandMessageMapper
MessageDispatcher
ResponseMessageBuilder
ServiceCommand
2. Map HTTP Request to CommandMessage
3. Build CommandMessage
4. Dispatch CommandMessage to INPUT Queue
5. Parse and extract CommandMessage
6. Map CommandMessage to ServiceCommand
7. Execute ServiceCommand
8. Build ResponseMessage
9. Dispatch ResponseMessage to OUTPUT Queue
10. Consume ResponseMessage
11. Parse Response
SignalR
SignalR
1. Send REST HTTP Request
12. Return Response
WEB #1
APP #1
APP #2
HTTP
To downstream
apps

14. Consistent Message Delivery
SimpleMath
Microservice
Inbound Message Queue
Outbound Message Queue
Inbound Messages
Outbound Messages
1 2
3
4
5
ASP.NET App

15. Consistent Message Delivery
SimpleMath
Microservice
Inbound Message Queue
Inbound Messages
ASP.NET App
1 2
3
4
4
4
5

16. Consistent Message Delivery
SimpleMath
Microservice
Inbound Message Queue
Inbound Messages
ASP.NET App
1
3
Queue Pool
2
45
6
Queue Pool Queue

17. Load Balancing a RabbitMQ Cluster
Node #1 Node #2 Node #3
Load Balancer

18. Load Balancing a RabbitMQ Cluster
Node #1 Node #2 Node #3
Load Balancer
Client Application
1
2
3

19. Load Balancing a RabbitMQ Cluster
Node #1 Node #2 Node #3
Load Balancer
Client Application
NewQueue1 NewQueue2 NewQueue3

20. Load Balancing a RabbitMQ Cluster
Node #1 Node #2 Node #3
Client ApplicationRandomiser

21. Load Balancing a RabbitMQ Cluster
Node #1 Node #2 Node #3
Client Application
Monitor Service
Queue Metadata StoreRandomiser

22. Selling to the Business
Real-world example of how to sell Microservices to techies and non-
techies alike.

23. Monolithic Flow #1
I want to book a flight

24. Monolithic Flow #2
Here are the available flights

25. Monolithic Flow #3
I want to book one of those flights

26. Monolithic Flow #4
Would you like to hire a car?

27. Business Drawbacks
 One-way communication
 Customer is in control
 Website is idle when user is idle
 Limited window of opportunity to interact
 Reduced scope for ancillary revenue

28. Technical Drawbacks
 Results in dependencies
 Failure affects everything
 Change is slow
 Scale is expensive (minor features require unilateral scale)
 Steep learning curve
 Technology stack is limited to specific skillsets
 Introduces legal pitfalls (PCI DSS, Compliance)
 Duplicated components due to lack of explicit boundaries
 Rigid – likely to break under pressure

29. Microservice Flow #1
Car Hire
Hotel
Flight
Taxi
User Profile
Fare Finder
Flights
Other

30. Microservice Flow #2
Car Hire
Hotel
Flight
Taxi
User Profile
Fare Finder
Flights
Taxis
Other

31. Microservice Flow #3
Car Hire
Hotel
Flight
Taxi
User Profile
Fare Finder
Flights
Taxis
Cars
Other

32. Microservice Flow #4
Car Hire
Hotel
Flight
Taxi
User Profile
Fare Finder
Flights
Taxis
Cars
Hotels
Other

33. Enhanced Flow Step #5
Car Hire
Hotel
Flight
Taxi
User Profile
Fare Finder
Flights
Taxis
Cars
Fares
Hotels
Other

34. Business Benefits
 Two-way communication
 We’re in control (think Google)
 APIs are always working
 Unlimited opportunities to interact
 Broader scope for ancillary revenue

35. Technical Benefits
 Eliminates dependencies
 Failure is isolated
 React to change quicker
 Scale is less expensive (APIs scale individually)
 More intuitive learning curve
 Technology stack is not limited to specific skillsets
 Shielded from legal pitfalls
 Reusable components
 Flexible – will bend rather than break under pressure

36. General Advice
 Log everything
 Log everything
 Log some more
 If a framework exists, use it
 Try to avoid the technical swamp

37. Let’s Talk Further…
 insidethecpu.com (search for “Microservices”)
 https://github.com/daishisystems/Daishi.AMQP
 @daishisystems
 https://ie.linkedin.com/in/daishisystems
 Q&A