What is Event Driven Architecture (EDA)?
2 min read
Event-driven architecture is a software architecture and model for designing applications.
MicroService Architecture (MSA) is the architecture attracting the most attention in Enterprise IT, with advantages such as fast deployment cycles and multi-language programming based on loose coupling between services.
In addition, it is developing rapidly as it fits well with container-based platforms (K8s, Docker).
However, MSA has a new requirement: Database Per Service. This makes it difficult to manage transactions across the separate DBs of distributed services, and we also need to take care of synchronizing semi-normalized data.
To supplement this, we must apply Event Driven Architecture.
What is Event Driven Architecture?
Event Driven is a very old keyword in the IT space.
Event Driven Architecture (EDA) is an architecture that stores an event log when an event occurs in a system where MSA is applied, operates based on it, and performs integration within the system through asynchronous communication.

Event Driven Architecture
When an event occurs in one service, it is propagated to other services in near real time, with each service executing an action in response to the event.
You can publish events in a number of ways. For example, you can publish events to a queue that delivers them to the appropriate consumers, or you can publish events to a publish-subscribe (Pub-Sub) message model that allows access to all stakeholders.
ActiveMQ and RabbitMQ are representative queue processing methods, and Apache Kafka is typically used as a message model.
Event
An event is a change of data (update, insert, delete).
ex.) Order creation, product purchase, user withdrawal
When using this architecture
- When multiple subsystems need to handle the same event
- There is no branching or branch consolidation, and new consumers can easily be added to the system
- Subsystems can independently check event streams
- Consumers can respond as soon as the event arrives
Asynchronous or Synchronous processing
Let me explain with a simple example what happens if we use synchronous processing (like rest) in MSA.
Let’s say a user places an order for a product.
- First, it reduces the inventory in the product Service
- Then it creates an order
- It sends the order information to the seller
- …
If the service grows, more processing has to be added to all of this logic.

Asynchronous processing
But if you use an asynchronous message queue, the producer just needs to create an event. Then the services interested in the event (OrderService, SellerService, ProductService) will receive it.
This can reduce coupling between services and simplify the flow.
Concluding
EDA is a concept that meets the new requirements arising from MSA architectures. It effectively responds to service and DB separation. EDA helps to troubleshoot and overcome the limits of MSA architectures through fault-tolerant distributed architectures, event brokers, and more.