To supply various consumers with price data, an event-driven data mesh is being designed and implemented in the Google Cloud. In the course of the modernization of the application landscape, the current data storage (IBM DB2 z/OS) is replaced by a new decoupled data storage in the cloud. The architecture of the application is considered a model for the replacement of other central data sources. PTA supports the development of the necessary infrastructure as well as the implementation of the required application components.
To replace the data storage, which has been based on a central IBM DB2 z/OS to date, a solution is being set up in the Google Cloud. Via third-party software, transactions on the DB2 are transferred to an event stream for incoming asynchronous events after the commit. For traceability within the targeted event-driven system architecture, all events are recorded in persistent event repositories. The event repositories are built and integrated with Google Cloud's own NoSQL product Firestore. Through event-driven triggers, further steps in the supply are forwarded to services in Cloud Run via various Cloud Functions. From now on, these serve as domain services and are responsible for the new persistent data storage, based on Cloud SQL. Subsequently, changes to the new persistent data model are published by another outbound event repository to an outbound asynchronous interface based on PubSub.
An event-driven data mesh represents a future-oriented architectural concept for the distribution of data. It enables the development of event-driven applications that can process business transactions in real time. For the business departments, this means the possibility of faster throughput times for operational processes.