Handling physical assets is slower than with virtual objects by its very own nature.
Working with cloud/virtual assets is way easier than with actual hardware and OT systems. That is my personal opinion, but I've got some reasons to back it up.
We can manage cloud/virtual resources like software. That abstract approach enables fast development cycles utilizing:
* Infrastructure as Code (IaC) capabilities
* automated deployment and testing
* unlimited number of development/test environments
Copy of the production environment needs minutes to deploy. We can experiment/break/destroy/recreate it as frequently as necessary, doing no harm to the business continuity. That is a pretty fantastic capability!
Some companies try to implement this approach in the hardware space, but handling physical assets is slower than with virtual objects by its very own nature.
That is why I do claim that working with cloud/virtual assets is easy.
The real question is how to make developing hardware and OT infrastructure as agile as software.
Any ideas?
12 Replies
This is a really interesting idea and one I've been exploring a lot in relation to Ethernet TSN.
IMO, the ability to have a unified deterministic network increases the portability of application workloads, driving the need for higher performance multicore processors further out on the edge. If used in a control environment, any multicore processor is likely running multiple workloads and therefore probably using embedded virtualization or a hypervisor.
The hypervisor or VMM or even an RTOS can manage those workloads, and open the door for real-time or near-real-time control software running in a partition like a vPLC: https://www.siemens.com/global/en/products/automation/systems/industrial/plc/simatic-s7-1500/virtual-plc.html.
In this context, the edge is more like a data center and the hardware becomes an abstracted resource. You can move software workloads around to wherever the resource is available, so "developing the hardware" isn't as important.
What do you think?
I am not worried about managing the software at the edge gateway and executing workloads. That scope belongs to the IT part from my perspective. The challenge is to increase the agility of OT systems (actual hardware, production lines, PLCs, etc.). We can not easily experiment in the OT area because working with physical assets is more complicated than operating on virtual abstractions.
Doesn't converting the bulk of edge infrastructure to Wintel/Lintel or similar address some or most of this?
If you can abstract what used to be a physical edge hardware function into software, haven't you accomplished what you're referring to?
Nope, a legacy production line in a factory is not easily "abstractable". Installing a server won't solve those challenges.
This seems like one of the big early use cases of technology like virtual PLC. Install it in parallel or in place of an existing controller. You may not be able to rip and replace the entire production line in one fell swoop, but you can evolve certain components and subsystems over time.
Of course if you're talking about the entire production line in general as one large logical system, then yes that's something different.
Yes, I am thinking about the whole thing.
Yeah, then digital twins or models I guess are the closest thing we have today
But that's not the same as what you're referring to
@dougsandy_92230 check this out
I have been working within PICMG/DMTF organizations to create low-level composable data models for abstracting factory OT technology with a goal of enabling digital twinning that encompasses the physical phenomenon of production. I believe that twinning of the physical phenomenon and their interrelationship with factory operations (the entire line) is the only way to realize desired goals of improved quality and higher yields. @LMtx it sounds like we are trying to tackle the same or similar problems. How do you see this playing out?
Two years ago, I created this video to explain the concept of digital twin: https://www.youtube.com/watch?v=FmjdKw1ZkO4&t=223s
LMTX DEV
YouTube
The concept of a Digital Twin in the Internet of Things
In this video I explain the concept of an IoT Digital Twin using few examples - from digital twin of a simple sensor to complete factory (Industrial Internet of Things).
I will cover what is needed to provide business value from the Digital Twin and how to extend this concept.
You can find additional details about this topic on my blog: https:...
That video is based on my project engagement and represents my approach during the actual initiative.
I often start by educating various stakeholders on the customer's side before talking about technical implementation. It helps to build a shared understanding and designing the roadmap.
Who are the technical stakeholders in these discussions, LMtx? Something I struggled with as a journalist was conveying why a developer/development manager should care about digital twins today and how they can prepare for it/leverage it in their work now. Your description here is the best I've seen that isn't completely abstract.
The stakeholders were spread across the field - from security administrators and shop-floor operators to high-level business owners. The CTO decided to start the digital transformation project, and executives needed someone to help make it happen 😉