Introduction to CloudOps for Azure

Elastic Path CloudOps for Azure is Elastic Path’s automated approach to run Elastic Path Commerce in a developer ready fashion. It utilizes the Microsoft Azure Kubernetes Service (AKS), which runs Docker containers in a Kubernetes cluster supported by the Azure Database for MySQL. Helm deploys and manages the cloud native services.

By utilizing CloudOps for Azure, users are able to have a stock version of the Elastic Path Commerce solution running in Azure in a short period of time. With Kubernetes’ tooling, developers can customize and deploy developer ready stacks where they can debug code running in Azure. Given the broad adoption and standardization on Kubernetes, Ops teams have a solution to which they trust and that adapts to their needs.

Build and Deployment Features

CloudOps for Azure provides a way to build and run a developer-ready deployment of the Elastic Path Commerce stack. Users can also run unit tests with the ep-commerce repository provided. With CloudOps for Azure, you can:

1. Build a deployment package from the ep-commerce source
2. Convert the deployment package into a set of Docker images
3. Create a Docker repository for the images using ACR, Azure Container Registry, service
4. Bootstrap an auto-scaling, load balancing Kubernetes cluster using Azure’s AKS service and HAProxy based Ingress Controllers
5. Populate an Azure Database for MySQL server
6. Create Kubernetes ConfigMaps and Deployments inside a Kubernetes cluster for running the Elastic Path stack

Note: Elastic Path CloudOps for Azure does not test any of the artifacts beyond the provided ep-commerce unit tests, do production-grade monitoring, or logging infrastructure for the deployment.

Production Usage

For users evaluating CloudOps for Azure, a best practice for all deployments is to ensure that the scaling and upgrading nodes in an AKS cluster provisioned/managed by Microsoft is thoroughly tested. As a result, it is encouraged for anyone evaluating CloudsOps for Azure to:

1. Thoroughly test scaling of the AKS cluster in different production-like scenarios
2. Contact Microsoft to inquire about the progress on resolving connectivity issues for the Kubernetes kube API and for kube-dns

For more information, see the following issues in Azure Kubernetes Services:

• Pod fails during cluster scaling
• DNS: Intermittent failure to establish new connection

Note: CloudOps for Azure deploys a cluster-autoscaler, which is the standard for autoscaling virtual machines in a Kubernetes cluster. Microsoft shows cluster-autoscaler as being in a Preview state. For production workloads, remove cluster-autoscaler and manually scale the virtual machines in an AKS cluster until cluster-autoscaler is no longer in Preview.