Design principles for an Event Driven Architecture

According to Wikipedia, an Event-driven Architecture, is a software architecture pattern that promotes the production, detection, consumption of, and reaction to events.

There is a perfect pairing between microservice-based architectures, Domain Driven Design (DDD) and event-driven architectures. In this conference we will review what design principles are the catalyst for this symbiosis as well as practical examples in different areas including governance. Many business use cases can be articulated on top of these principles, abstracting them from both complexity and variability in the technological stack.

As a good part of the audience will already be dealing with events and microservices, we will also explain other key concepts:

* Designing a future-proof event taxonomy.
* Strategies for event enrichment, starting with the definition of that concept.
* Managing correlation or inference of events.
* Benefits from an event schema registry using for example Apache Avro.
* Traceability of events by design.
* Data conciliation patterns, and when to avoid it.

We will also take advantage of the opportunity to discuss about common challenges (and others not that common), frequent mistakes and how to avoid or mitigate them.

To conclude, we will explain some use cases that we are solving superbly based on real-time events: Communications, Order Management, Business Activity Monitoring (BAM), KYC, GDPR, ...

Transcript

1. Design principles for an event driven architecture Madrid Apache Kafka Meetup Luis García & Pablo Ruiz - ING October 2019
2. Introduction 2 There is a perfect pairing between microservice-based architectures, Domain Driven Design (DDD) and event-driven architectures
4. Event Taxonomy 4 Purpose Driven Technical Meaning Driven Business Domain Raw Native Standardized Atomic raw domain Derived (*) business
5. Taxonomy at microservice level 5 (*) each bus represents one specific topic in the Event Bus Domain API BB viewed from the outside Business Technical API Storage Bus Connector CORE Raw Domain Building Block internals Business Technical
6. Standard language 6 API Call API Call Ubiquitous Language Ubiquitous Language Ubiquitous Language Standard Language
7. Bounded Context Bounded Context Cohesion (strong) 7 Producer Consumer Schema Registry Send Schema Read Schema DB DB
8. Reference Data 8 Building Block A Building Block B Master Reference Data Building Block C Datalake
9. Coupling (loose) 9 Product A transaction received Fraud Engine transaction received Fraud Alert!! Communications transaction received Contact customer!! Channel Gateway Contact customer!!
10. Correlation & Inference 10 Domain A Event A Domain B Domain C Event B Domain A Event A Domain B Domain C A & B A & NOT B
11. Traceability & Lineage 11 Domain A Event Id: 1 TraceID: 1 Domain B Domain C Event Id: 2 TraceID: 1 Event Id: 3 TraceID: 1 Event Id: 4 TraceID: 1 Domain D Event Id: 5 TraceID: 5
12. Aggregation / enrichment 12 Domain A Domain B Domain C Name: Peter Surname: Parker Job: Photographer Name: Peter Surname: Parker Alias: Spiderman Domain D Name: Peter Surname: Parker
13. Orchestration 13 Orchestrator Generate Monthly Extract Product A Product B Document Mgm Product A movements Product B movements Statement request with Product A&B movements
14. Choreography 14 Product A transaction received Category Service transaction received transaction categorized Customer Reporting transaction received transaction categorized
15. Reconciliation 15 Domain A Domain B Domain C Domain D Reconciliation? Reconciliation?
16. Privacy / Data Segregation 16 Producer Clear Message <encoded message> Clear Message Producer Consumer Name: Clark Surname Kent Income: 1234.00 Name: Clark Surname Kent Income: 1234.00 Other Consumer Name: **** Surname ***** Income: 1234.00 <encoded message> Consumer Other Consumer <encoded message>
17. Technical/Business Monitoring 17 Domain A Domain B Domain C Alerting
18. How EDA impacts a microservice based architecture 18 Feature EDA impact Loosely coupled Location transparency Strong cohesion Organized around business capabilities Each service is independently deployable Highly maintainable – Easy to maintain Highly testable – Easy to test Each service is as simple as possible Data Management Transaction Management
19. Thank you!! Questions? 19