David Aronchick, Head of Open Source Machine Learning Strategy, Microsoft
- The attack surface your device offers is potentially enormous. In planning for system security, know what you want to allow the device to receive and how it can take action.
- An important architectural consideration that influences the nature of device communications across the control plane is how much processing occurs locally on the device and how much occurs in the cloud.
“If you don’t have a control plane that can look across all your devices, . . . you’re going to be in for some bad times.”
The number one success factor in Internet of Things solutions is having an update strategy to remotely monitor, control, and manage all devices. So maintains David Arnonchick, head of Open Source Machine Learning Strategy at Microsoft. “If you don’t have a control plane that can look across all your devices, help you deploy them successfully, maintain awareness of what’s going on, allow you to update them reliably, and take offline those components that are no longer relevant, you’re going to be in for some bad times,” he says.
That control plane plays a critical role in everything from deployment and life-cycle management to normal device operations and even system and device-level security. “That universal control plane enables you to fix many problems down the line,” says Aronchick. So, how do you create it?
To do that, you have to think architecturally about the device and the solution you are designing, which in turn requires answering some fundamental questions. For example, how regularly will your device communicate, and how reliable will those communications be? Will the device be deployed in high-bandwidth or low-bandwidth environments? Does it need to talk to a central server, or can you have hubs in your environment to accommodate the local network and local updating?
