Lynx teams to put cloud AI containers at the edge

Lynx teams to put cloud AI containers at the edge

Technology News |
By Nick Flaherty

Lynx Software Technologies has worked with Google to run containers of AI software at the edge in mission critical industrial networks.

Lynx has ported the Google Anthos framework to its MOSA.ic platform. This allows containerized software services from the cloud such as Google Cloud Visual Inspection AI service to run securely as bare metal applications.

Related articles 

“This partnership also marks an important step in our growing industrial ecosystem,” said said Pavan Singh, Vice President of Product Management at Lynx. “The partnership is to extend Kubernetes to deploy a Kubernetes cluster at the edge, deployed on Anthos in one box. Now a model that is built in the cloud can be deployed as a container on that Kubernetes cluster and right next to that cluster you have the fusion and data ingestion happening. That feeds data to this cloud connected ML workload,” he said. “What we are able to do is isolate the cloud connected workload and allow the data to feed to the container without opening up the system to all kinds of zero day attacks.”

Lynx has used its separation kernel technology to provide a validated solution for secure, video-based quality inspection in industrial and energy facilities.

This allows easy deployment of real-time image capturing on devices such as cameras on manufacturing plant floors and inference models built by Google Cloud Visual Inspection AI in Kubernetes containers that generate the insights. These can run securely alongside the supervisory controller that connects to the Manufacturing Execution System (MES) without compromising the security of the network.

This provides the three key functions of image capture from the camera, the inference engine on Google Anthos and the supervisory controller are completely sandboxed with the option of secure one way (data diode) connections between them. This is handled via an inter-VM messaging framework using shared memory.

“Running on Mosa.ic creates multiple bare metal partitions so to Google Anthos it looks like its running on bare metal. Now you don’t need an expensive, power-hungry box, it can be on the factory floor or on an oil pipeline,” said Singh. “Mosa.ic is built on a separation kernel, taking the kernel and exposing that as a partition. We are separating out the networks then also separating the compute and memory at boot up so each workload only knows the compute and memory that is assigned to it.”

The first version is optimized for image analysis and runs on an Advantech MIC770 which uses the Intel 8th Gen Core i CPU socket-type (LGA1151) with the Intel Q370/H310 chipset. However, Lynx says MOSA.ic is able to run on various Intel and ARM processors.

Other related articles

Other articles on eeNews Europe 

If you enjoyed this article, you will like the following ones: don't miss them by subscribing to :    eeNews on Google News


Linked Articles