Mark Glover spoke to Roy Chikballapur about the role of edge computing in Industry 4.0, how a comparison with the cloud is not useful and why bringing the two technologies together could benefit the smart factory.
In a recent LinkedIn post I pondered if edge computing would ever replace the cloud.
service innovation from Automation
A comment from Roy Chikballapur from MachIQ caught my eye. “I would say that they are heavily complementary, and neither can deliver without enough advancement of the other,” he wrote, “The edge hyperbole is an automation-industry led backlash that doesn’t fully capture the opportunity of the speed with which innovation can be implemented on the cloud and distributed to the edge.”
This was all great stuff. I quickly messaged Roy and we arranged an interview the following day. This was the kind of insight I was hoping for, however the Skype call we set up had other ideas, booting Roy off the line halfway through.
Skype operates via the cloud, having moved there in 2016 after existing primarily in P2P networks. However, as proven by my call with Roy, connections can drop out as the application gobbles up bandwidth.
Our connection used a trickle of data compared to that of a connected smart factory and it seemed ironic that our video-chat set-up to discuss the role of cloud and edge computing should drop out.
“You can’t have (connection issues) like we just had right now,” Roy says, using our dis-connection as an example of what can’t happen in an industry setting. “You can’t afford that break in the network when you’re running any type of factory because machines just can’t stop.”
Roy is absolutely right. For manufacturers time is money and asset downtime is loss making. The digital sinews that connect smart factory assets need to work all the time and Roy’s initial comment in response to my LinkedIn post, that the cloud is unable to handle these intricacies (that it was unable to handle our Skype call is another article) made sense.
Of course, the cloud is not in the sky, it’s very much on the ground, stored on data centres and server farms, sometimes a long way from the asset that’s asking to communicate with it. It’s this geographical distance that causes ‘lag’ or latency in “cloud speak”- the time it takes to get the data sent, dealt with and then sent back. A lag of several seconds may seem trivial but for industry it’s pivotal.
The news item that prompted this piece, centred on an analyst report from GlobalData who suggested edge computing was primed for the Asia Pacific (APAC) region.
"Edge computing operates by being physically closer to a device..."
The APAC consists of East and South Asia, South East Asia and Oceania and is made up of large sprawling countries like China and Australia. It covers approximately 2.8 billion hectares of land blanketing nearly 22% of the global land area. It’s huge.
In the report, GlobalData predicted benefits for the region’s manufacturing sector if it fully adopted the edge including the streamlining of industrial processes, supply chain improvement and more autonomous use of equipment.
The uptake, the report claims, is due to the widespread adoption of connected assets in the region including smart homes. It forecasts by 2024, given the current adoption rate, it will become the second-largest market of edge computing, second only to North America.
In contrast to the cloud, edge computing operates by being physically closer to a device, rather than stored in a data centre several hundred miles away; operating on the ‘edge’ of a network thus transporting the digital instructions quicker and negating lag.
The report says China will spearhead the adoption, and given the country’s size (9.5 million km2), lag from a server in a Beijing datacentre for example to an end user in Shanghai could be avoided.
If, as the report claims, manufacturing is to benefit, it’s important to understand Industry 4.0; the fourth industrial revolution as it were, where data is the new coal and a natural environment for the edge to operate.
In our home, we tinker with the heating while on the train home from work; or look through our security camera while waiting for a bus. It’s this societal shift in the use of connected devices, Roy says, that has pushed the fourth industrial revolution forwards.
“If you see where the Internet of Things came from, it has origins in implementations such as Google’s Nest,” he explains. “Here the sensors would send signals to the Cloud to make sense of the data, determine what action the device needed and carry them out.”
This was a step on from Programmable Logic Controllers (PLCs), the first industry-based conduit for connecting machines to networks and a significant step in manufacturing’s development.
Pre-internet and pre-big data their use was limited but large B2C tech firms: Microsoft and, in this instance, Google, with the eventual advent of the cloud, took the technology further. Today can get your living room lights talking with your mobile phone. “That was basically how Nest really started,” Roy says.
"The industrial firms knew that the cloud alone could not sustain the real-time demands of smart factory operations..."
PLCs were game changers for of GE and Schneider in terms of industrial automation (Schneider would eventually purchase Modicon, the company credited with producing the first PLC), they understood the value of real-time data in a factory setting; that continuous uptime and production was paramount to their operations. Which bring us back to Roy’s point on industrial continuity and how the lag from cloud connectivity will not only affect output but, most importantly, employees’ wellbeing too.
“If it takes multiple seconds for data to travel back and forth, for instructions to be sent from a cloud-based controller to a device in the factory,” Roy says, “it can sometimes literally mean the difference between life and death as there are cases when worker safety inside the plant is affected. And that just can’t happen.”
The edge vs cloud debate was fuelled by industrial actors like GE, Schneider and Siemens responding to IT players including the aforementioned Google, Amazon and Microsoft’s assertion that the cloud could oversee everything, including industrial, operational technology.
The industrial firms knew that the cloud alone could not sustain the real-time demands of smart factory operations. It just wasn’t possible. The stand-off went on for some time until Moore’s law – his 1965 theory that over time, more and more transistors would fit into silicon chips - made it economical to embed sufficient data storage and processing capacity within the asset itself, enabling much of the real-time computing that make smart factories “smart”.
“Computing that was initially supposed to be carried out in the cloud could eventually be carried out on the device itself, by attaching a processor inside. This is basically what edge computing is.” Roy says before suggesting the technology could and should work in perfect harmony with the cloud, particularly in industrial environments. ”There is a space for both these technologies to operate side by side and complement each other perfectly well, to deliver exponentially higher value to industrial plants than either one being utilised alone.
There’s more to discuss here, particularly around edge and the Cloud working together, but I’m saving this for an episode of the Field Service Podcast with Roy as my guest.