Solution accompanying the Linkedin article on Serverless architectures - covers a scenario where an Organization implements a Business to Business Integration with 2 Trading Partner Organizations.
The Trading Partners upload the orders to an AWS S3 Bucket. For simplicity here, a single AWS Account is used with 2 S3 Buckets created in them, one per Trading Partner. Per Partner there is an ADF Pipeline that picks up the order data from S3 and inserts them to a designated container in a common Azure Blob Store Account svlsb2bin. Screenshot below shows the ADF Connection to an AWS S3 Bucket.
The ADF Pipelines accompanying this article are available in the ARM Template ./arm template/datafactory/arm_template.json. Run this template to deploy the ADF Pipelines to your subscription. Edit the arm_template_parameters.json file to add information like the AWS S3, Azure Storage accounts and Access keys applicable to your Subscription.
Upload the sample files to S3 Buckets, trigger the ADF Pipeline in Azure to move the files to Azure Storage Account. Screenshot below shows the outcome of the Pipeline run. Sample files for order upload are located in the folder ./sampleorders/().
From the Settings pane of the Azure Storage Account svlsb2bin, an Event Grid Subscription is created for the 'Blob created' Event on the Logic App ordergenflow. When an order is inserted in the Storage account, it triggers the Logic App process. Screenshot below shows the Logic App trigger configuration for Event Grid.
The Azure Logic App accompanying this article is available as a Visual Studio Solution in this Repository located at ./b2blogicapp/b2blogicapp.sln. Compile the Solution, right click the Logicapp.json file to view it in the Designer.
To deploy this Logic App to your Subscription, select 'publish' on the Logicapp.json. Edit the configuration parameters in it to suit your Azure Subscription, before publishing.
For simplicity here, a single Logic App is used to handle the process flow pertaining to both the Trading Partners. Based on the folder where the order document was inserted, the Logic App instance determines which Partner this order document pertains to. It then branches into a Partner specific flow. The incoming data is parsed using a JSON Schema, and Business rules are applied to determine whether the order can be approved, or whether it should be flagged for user intervention. The Business rules in the flow are, for simplicity, embedded in the Logic App itself, rather than looking up an external database. After the Business rules are applied, the order status is updated. See screenshot below which shows how the Order status is updated based on the outcome of a Business rule evaluation.
Since the order data schema is different across Partners, before the data can be saved in the Azure Storage Table ordersdata, it has to be mapped to the target schema. For simplicity here, the source document itself is modified by the Logic App using 'addProperties' and 'removeProperties' inline functions. For more complex and production scenarios, use the transform activity in Logic Apps to map and transform the document to a target schema.
The Partnername in the order is used as the Partition Key in the Storage Table, and the Order number is used as the Rowkey At the end of the Logic App flow, the order data gets stored in Azure Storage Table.
The Visual Studio 2017 Solution of the Azure Function App ordersflow is available in the Respository at ./ordersflow/ordersflow.sln. There are 2 Functions in it that are exposed as HTTP callable APIs
- Get orders pending approval
- approve an order in the Storage Table
To deploy this Function App to your Subscription, edit the localsettings.json file to update the Connection string used to connect to the Azure Storage Table. Publish the Project to your Azure Subscription.
Acts as the facade to the APIs exposed by the Function App. A Consumption tier is chosen for the API Management Gateway instance for the Serverless capabilities. From the API Management Console, use the built-in ability to import an existing Function App. From the testing console on the API Management Console, ensure you are able to execute the APIs exposed by the Function App successfully.
From the API Management Gateway instance orderfacade, use the built in feature to export the Open API specification for an API. Import this file into PowerApps to create a custom connector. See the screenshot below. Test each of these definitions and ensure that you are able to invoke the Function App APIs. At the end of this step, a custom connection instance is created from this custom connector in PowerApps.
The PowerApps App OrderApproval file accompanying this article can found here - ./powerapp/OrderApproval.mapp
Load all the orders pending approval, and ensure that you are able to update the status to 'approved' or 'rejected' as the case may be.
The ASP.NET Core 2.0 Mvc Project is available here ./Portal/OrderFlowBusinessMonitor/OrderFlowBusinessMonitor.sln
On the Visual Studio 2017 project, Docker Containers is enabled, and Docker Swarm is selected as the orchestrator. This creates the necessary Dockerfile & Docker Compose files for the Solution. Use Docker commands to tag and push the container image to Azure Container Registry, after compiling the Solution.
This Web Application is integrated with Azure Active Directory for user authentication, using the turnkey capabilities in the Visual Studio 2017 Project Template. The option to support Multi-organization Azure AD Tenant was selected when creating the Project to support Multi-tenancy.
To test the scenario, 2 Azure AD Tenants were created in the Azure Subscription, one per Trading Partner. A user in the Role 'Global Administrator' was created who would approve the use of the Web Application by all Organiational users in that Azure AD Tenant.
An ARM Template, available at ./arm templates/sfmesh/deploysfmeshapp.json, is used to provision the Azure Service Mesh Application. Deploy the Service Fabric Mesh Application using this Template. Retrieve the Public IP Address created for it during the deployment operation to launch the Application. On user sign in, the order data should be visible.
Screenshot below shows the list of Azure Services deployed to implement this scenario. Use this URL http://104.45.72.188:50039/ to access the application, and use the following credentials
Trading Partner 1: user name : johndoe@neocorpone.onmicrosoft.com pwd: Orderb2b@987
Trading Partner 2: user name: ravikumar@femacorpone.onmicrosoft.com pwd: Orderb2b@987
