Overview

This is a Proof of concept of a CQRS implementation using Azure Services.

This is an improvement over this project, the different components have been deployed on Azure using the relevant services.

Technical Overview

This is the full architecture :

Full presentation can be found here : Presentation link

Reads are happening through Elasticsearch, Writes happen through CosmosDB. Synchronization is done using an Azure function that subscribes to an Event Grid Topic.

The whole architecture is deployed using Terraform.

We used Azure Spring Apps to deploy the different microservices.

We used a Redis cache to cache Responses from Elasticsearch for even faster queries.

We used Application Insights Both in the Azure function and Spring apps service to monitor the state of our set of applications.

The Configuration Server and the Registry Server are automatically provided by Azure. We just configured the configuration server to point to the appropriate git repository. You can find the configuration server repository here .

Quickstart

1. Create terraform infrastructure by running terraform apply under /infrastructure.

2. deploy the azure function ( after building it with npm run build in /function-app ).

3. Create the event subscription in the terraform code, to link the azure function to event grid.

4. change Eventgrid endpoint + Token and cosmos connection string in the config server (on github) to match the new ones ( you can find values in the terraform output ).

5. in each of the services ( search, upsert, gateway ), we package them again :

./mvnw clean package -DskipTests

6. We deploy them on Azure spring Apps :

az spring app deploy     --resource-group dev-cqrs-rg     --service dev-cqrs-springcloud     --name search-microservice     --artifact-path target/search-microservice-0.0.1-SNAPSHOT.jar --deployment default

az spring app deploy     --resource-group dev-cqrs-rg     --service dev-cqrs-springcloud     --name upsert-microservice     --artifact-path target/upsert-microservice-0.0.1-SNAPSHOT.jar --deployment default

az spring app deploy     --resource-group dev-cqrs-rg     --service dev-cqrs-springcloud     --name gateway-service    --artifact-path target/gateway-service-0.0.1-SNAPSHOT.jar --deployment default

Possible Improvements

• The provisioning is done by Terraform. However some configuration is still not automated ( Updating the different uris/endpoints in the config server, Bundling/Deploying the Azure functions & the different spring boot applications ). For this we can create a CICD pipeline to automate the process.

• The configuration server secrets should be secured more in a real setup. For that we can use Azure Keyvault as a backend for our spring configuration server.

• The microservices lack proper unit tests which is very important in a real setup. In this simple example this wasn't necessary.