Loosely Coupled Asynchronous Microservices

Anti-Pattern - Tightly Coupled Synchronous Microservices

The latency is the sum of all calls. If the typical call latency is 50ms, in the above example the total latency is 250ms.

The reliability is the product of all calls. If the reliability of one service is 99.5%, in the above example the overall reliability is 97.5%.

In order to improve performance and reliability the number of synchronous service-to-service calls needs to be minimised.

Services should be as self-contained as possible.

Loosely Coupled Asynchronous Microservices

Services publish Domain Events on Kafka containing data that may be of interest to other services.

A Service listens for Domain Events on Kafka containing the data it requires and maintains a local copy.

Service A has all the data locally and works even when downstream services are unavailable (although, some of the data may be out of date).

A call from a channel has 50ms latency and 99.5% availability - 5x better than the synchronous example above.