Hands-On Microservices with Kubernetes

Applying the saga pattern to microservices

Relational databases can provide ACID compliance for distributed systems through algorithms, such as two-phase commit and control over all the data. The two-phase commit algorithm works in two phases: prepare and commit. However, the services that participate in the distributed transaction must share the same database. That doesn't work for microservices that manage their own databases.

Enter the saga pattern. The basic idea of the saga pattern is that there is centralized management of the operations across all the microservices and that, for each operation, there is a compensating operation that will be executed if, for some reason, the entire transaction can't be completed. This achieves the atomicity ...