Microsoft to better meet the needs of adopting solutions that can extend your environment, from the main datacenter to the peripheral sites, with innovative Azure services, makes the Azure Stack portfolio available to its customers. It is a set of hybryd cloud solutions, that allow you to deploy and run your application workloads consistently, without restrictions imposed by the geographical location. This article provides an overview of the Azure Stack Edge platform (ASE) and its characteristics, examining the use cases and the main features.
Before going into the specifics of Azure Stack Edge it is good to specify that the solutions included in the Azure Stack portfolio are the following:
- Azure Stack Edge: the Azure managed appliance that can bring computational power, cloud storage and intelligence in a remote edge of the customer.
- Azure Stack HCI: the solution that allows the execution of virtual machines and an easy connection to Azure thanks to a hyper-converged infrastructure (HCI).
- Azure Stack Hub: the offer for enterprise companies and public sector customers, needing a cloud environment but disconnected from the Internet, or need to meet specific regulatory and compliance requirements.
To get an overview of these solutions I invite you to read this article.
Azure Stack Edge value proposition
The results that can be obtained by adopting the Azure Stack Edge solution are the following:
- Possibility of adopting an on-premises model Infrastructure as a service (IaaS) for workloads on peripheral sites (edge), where both hardware and software are provided by Microsoft.
- Ability to run applications at customer sites, in order to keep them close to the data sources. Furthermore, allows you to run not only proprietary and third-party applications at the edge, but also to take advantage of different Azure services.
- Availability of built-in hardware accelerators that allow you to run machine learning and AI scenarios at the edge, right where the data is, without having to send data to the cloud for further analysis.
- Possibility of having an integrated cloud storage gateway that allows easy data transfer from the edge to the cloud environment.
The main scenarios for using Azure Stack Edge are the following:
- Machine learning at peripheral sites: thanks to the presence of integrated hardware accelerators and the processing capabilities offered by the solution, you have the ability to cope with these scenarios right where the data resides, processing them in real time, without having to send them to Azure.
- Computational capacity at edge: customers can run their business applications and IoT solutions at peripheral sites, without necessarily having to rely on constant connectivity to the cloud environment.
- Network transfer of data from the edge to the cloud: used in scenarios where you want to periodically transfer data from the edge to the cloud, for further analysis or storage purposes.
To support the different usage scenarios reported, vertically between industrial sectors, Azure Stack Edge is available in three separate form factors:
- Azure Stack Edge Pro, a 1U blade server with one or two GPUs.
- Azure Stack Edge Pro R, a rugged server with GPU, in a sturdy carrying case, complete with UPS and backup battery.
- Azure Stack Edge Mini R, a machine with a reduced form factor with a battery and a low weight (less than 3,5 Kg).
Azure Stack Edge "rugged" versions allow resistance to extreme environmental conditions, and battery-powered versions allow easy transport.
Azure Stack Edge stack software
The customer can place the Azure Stack Edge order and provisioning directly from the Azure portal, and then use the classic Azure management tools to monitor and perform updates. Hardware support is provided directly by Microsoft, that will replace the components in case of problems. There is no upfront cost to obtain this appliance, but the cost will be included monthly in the billing of Azure services. Since, once configured, any application running on Azure Stack Edge can be configured and deployed from the Azure portal, eliminates the need for IT staff in the edge location.
Azure Stack Edge Computational Capacity
The ability to offer computational capacity taken from the edges is one of the key features of Azure Stack Edge, which can be provided in one of the following ways:
- IoT Edge: the execution of containerized workloads distributed through the IoT hub has always been supported since the launch of Azure Stack Edge and continues to be so.
- Kubernetes: recently, support was introduced for the execution of containerized workloads in Kubernetes clusters running on Azure Stack Edge.
- Virtual machines: another way to run applications is by activating workloads on board virtual machines.
Kubernetes environment in Azure Stack Edge
Kubernetes is becoming the de facto standard for the execution and orchestration of containerized workloads, but those who know these environments, is aware of some of the operational challenges that can arise from managing a Kubernetes cluster. In this context, the goal of Azure Stack Edge is to simplify the deployment and management of Kubernetes clusters. With a simple configuration, you can activate a Kubernetes cluster on Azure Stack Edge.
Once the Kubernetes cluster has been configured, you must perform additional management steps, that are simplified in ASE with simple add-ons. Among these operations we find:
- The ability to easily enable hardware accelerators.
- The provisioning of the storage system to create persistent volumes.
- Keep it up to date with Kubernetes releases by taking the latest updates available.
- The ability to apply security and governance mechanisms from their own infrastructure.
Cluster environment configuration completed, Simple mechanisms are provided for deploying and managing workloads on the Kubernetes cluster, by using the following modes:
- Azure Arc: ASE comes with native integration with Azure Arc. With just a few steps you can enable Azure Arc, allowing applications to be distributed in the Kubernetes cluster directly from the Azure portal.
- IoT Hub: by enabling the IoT hub add-on it is possible to use it for the distribution of conteiners.
- Kubectl: finally supports the native way kubectl, typically used in disconnected environments or if you have an existing infrastructure that already integrates with this mode.
Virtual machines in Azure Stack Edge
Another variant to offer computational capacity at the edges is the activation of virtual machines. Azure Stack Edge allows you to host virtual machines, both Windows and Linux, offering the ability to deploy and manage these virtual machines directly from Azure or by acting locally.
One thing to consider is that Azure Stack Edge allows you to set up simpler network topologies than Azure or Azure Stack Hub.
Regarding the hardware acceleration features in Azure Stack Edge, these two variants are supported:
- GPU NVIDIA T4, fully integrated with the GPU stack
- Intel Movidius Visual Processing Unit (VPU), for AI and ML scenarios
Azure services that can be deployed in Azure Stack Edge
The number of services that can be activated in Azure Stack Edge is large, among those recently introduced we find:
- Live Video Analytics: a platform for creating video solutions and applications based on artificial intelligence, to carry out real-time insights using video streams.
- Spatial Analysis: a real-time computer vision module to analyze videos and understand people's movements in physical spaces. For example,, during the Covid period, many retail stores want to implement social distancing policies and may use a special analytics module to understand certain behavior based on videos shot in the store.
- Azure Monitor: this increases application performance and availability by collecting logs from containers and analyzing them.
In business realities, the adoption of totally cloud-based solutions does not always turn out to be a viable choice or the best of all, hybrid solutions often have to be adopted, which in any case include the possibility of using the innovations introduced by the cloud. Azure Stack Edge is a flexible and modern solution that allows you to meet your needs, even the most challenging ones, emerging for edge sites, without neglecting the potential offered by the public cloud.