This week, a startup called Nubix raised $2.7 million to complete a seed round of funding totaling $4.5 million. The story caught my eye because Nubix is building a form of containers designed for edge computing jobs. Like Balena (formerly known as Resin.io), Pixieom, and a few other companies, Nubix is trying to bring the cloud mindset to the industrial world.
That, however, is a tall order.
We’re still spilling buckets of digital ink discussing the IT and OT divide even as the folks in the OT world recognize that their bespoke and highly regimented way of building and maintaining operations holds back their digital transformation projects. I’m not casting blame on these people. Their own attention to detail and focus on reliability is an essential element for computing in the real world, and the IT folks are recognizing that as well.
But much like developers gravitated toward the cloud in the late 2000s, and then mobile in the mid-2010s, the flow of software and computing talent is migrating to the industrial, OT world, says Rachel Taylor, the CEO of Nubix. As a former enterprise software developer and then a business-focused employee at several big enterprise tech firms, she has seen the flow of computing talent over the years.
“The edge right now is full of potential for companies to start to deliver new services and solutions that aren’t even thought of today,” says Taylor, referring to the many embedded computers managing machines that exist. “All of that is what is locked up in the devices: It is a massive amount of computing and the ability to unlock the services today requires new solutions.”
That’s what the three-year-old startup wants to deliver with its two-pronged approach to managing and building applications for the edge. Its product has a container-style software that will run on devices and then a web-based management platform on which developers can build applications for the edge and then deploy them.
The edge software is designed to run on the real-time operating systems that many of these tiny sensors and management computers use. Taylor says the Nubix container can run on Linux, but it’s really designed for smaller MCUs. So far, the company supports Amazon’s FreeRTOS, Microsoft’s Azure RTOS, and ARM’s Mbed, but can add more without too much trouble.
Competitors in the space are also building custom containers designed for edge devices, although in some cases those products use traditional Docker or Kubernetes containers. The thinking is that this will help the many cloud developers out there easily adapt to developing applications for the industrial edge.
Taylor’s approach is different. Nubix keeps the architecture design of a container, where developers can write an application once and run it in the cloud, on a gateway or a smaller edge device, but changes out the actual container for one designed with industrial IoT needs in mind. She says that failover isn’t as essential at the edge compared with the cloud. Instead, the focus has to be on fixing the downed device.
For example, if a temperature sensor fails and stops sending data, usually there is another one nearby that will continue to send data. Later, after the process using the data is complete, someone can fix the problem and get that sensor up and running. Spending lines of code on device failover in that example doesn’t make sense because there’s nothing to fail over to, and the focus needs to be on getting the original device back up.
This is made possible by the fact that in most industrial designs, the focus is on having physical redundancy of parts and sensors associated with important operations. Indeed, failover in the industrial world is based on fixing physical devices as opposed to spinning up a virtual server.
The risk of the Nubix approach is that engineers and developers won’t want to learn a new container model, although Nubix has tried to design a user-friendly drag-and-drop menu for building out applications on the web-based platform.
Nubix is much smaller and younger than its competitors in this space. Currently, it has just three full-time employees and several proofs of concept, but no customers. And yet, I like both the big vision Taylor has and the edge-first approach she’s taking. She recognizes that many industrial customers aren’t excited about changing up their computing environments, but knows they have little choice as older engineers retire and newer engineers come in with very different skill sets and approaches.
COVID-19 is also pushing many industrial customers to transition more quickly to smarter factory floors — as they need to manage equipment remotely — without requiring staff to work together in close confines. Having a low-code approach designed with edge use cases and practices in mind has a real chance of success.