Introduction

As the Internet of Things (IoT) adoption continues to accelerate the volume of telemetry grows. Most organizations are collecting data everywhere, all the time, about everything. This increases the cost of data collection in addition to reducing the timeliness of insights and security. This is a problem that I’ve been focused on solving with clients for some time now.

Azure IoT Edge

Today at Build, Microsoft announced Azure IoT Edge. Sam George – Partner Directory of Azure IoT has a blog post describing the service:

Azure IoT Edge enables an ecosystem of Microsoft and third-party services at the edge to help you light up new IoT scenarios. Azure IoT Edge enables seamless deployment of cloud services such as: Azure Machine Learning, Azure Stream Analytics, Azure Functions, Artificial Intelligence, Cognitive Services, and Azure IoT Hub communication and device management features

The Azure IoT Edge site describes the services as:

Extend cloud intelligence to edge devices: Run artificial intelligence at the edge, Perform edge analytics, Enable real-time decisions.

I find these explanations a bit vague and somewhat confusing.

Vague considering Azure IoT Suite is described as serving the same purpose of deployment.

Vague in not understanding how this new service fits within Azure IoT Central the SaaS offering of Azure IoT Suite.

Confusing in that these Azure services would not run on the edge devices themselves.

Then there was this statement:

Build advanced analytics, machine learning, and artificial intelligence in the cloud and deploy to physical devices using IoT Edge.

I’m very interested in how over the air (OTA) deployments of advanced analytics will occur on the Microsoft platform. The best I can determine is that perhaps they are talking about using an Azure machine learning web services from the edge device.

What is Edge

Regarding the use of the word edge. The industry defines Edge Computing as:

Edge computing is pushing the frontier of computing applications, data, and services away from centralized nodes to the logical extremes of a network. It enables analytics and knowledge generation to occur at the source of the data. source: wikipedia

The edge of an IoT ecosystem is not the cloud or gateways as these are not the extremes of the network nor are they the source of the data. It’s the devices attached to sensors/actuators themselves that is the edge. As such, the marketing material from Microsoft conflates edge with gateways that connect them to the cloud.

GitHub Azure IoT Edge

Sam George blog post further stated that Azure IoT Edge is an extension of the Azure IoT Gateway SDK. If that SDK was extended to support device analytics, then there would be a notable commit to the GitHub project. If you look at the commit activity of that project there was not substantial commit since last November. So I am curious how we accomplish device analytics in a new way. Looking at the modules there is not one that strikes as device analytics ether.

Final Thoughts

It may be clear to Microsoft how Azure IoT Edge relates to Azure IoT Suite and Azure IoT Central, but it’s not to the community. The messaging of a roadmap needs to improve a bit.

The use of edge is a bit misplaced. I believe Microsoft recognizes that they are well positioned on the cloud, but have some exposure on technology that extends to gateways end edge devices themselves.

Categories: Article