Super-fast automated decisions at the edge are where value lies for enterprise
Very low latency 5G networks are on the way, and incremental improvements are likely to go on indefinitely. Wise enterprises – and those who provide them with services – will not be focused on the speed of the network’s round trip, but how they can both gain competitive advantage from it.
20 Nov 2020
Super-fast automated decisions at the edge are where value lies for enterprise
Sponsored by: VoltDB
One of the most exciting aspects of 5G is ultra-low latency and the new use cases it will enable. We are looking to these new applications to transform enterprises and bring crucial new revenue streams to telcos. We’re generally hazy about exactly how this new value will come about, although we’ve whipped ourselves up into something of a frenzy about the extraordinary potential of ecosystems and the edge, and automation and AI.
So what do enterprises want to bring to the edge? The first thing to recognize is that the value is not in how few milliseconds (ms) the network can achieve the round journey, but in combining it with a complete package of capability into ecosystems that deliver maximum value from the speed at which complex decisions are automated.
The owner of the platform that delivers that capability holds the keys to the digital kingdom. Telcos spend a lot of time talking about moving up the value chain beyond connectivity, this is a great opportunity.
It’s often hard to put exact figures on the benefits of low latency, so this example made quite an impression. Mobile operator Comviva (formerly Mahindra Comviva), has an application called MobiLytix which pulls data from call detail records and other data about customers to present the best-option offer(s) for them in under 10ms. Does the 10ms matter?
This resulted in a 253% improvement in offer acceptance for the operator and is a great demonstration of the hidden value in enabling sub-10ms decisions. When you start to think about it, it seems the rapid, data-driven decision capability in conjunction with super low latency networks could be applied all over the shop as enterprises eye the edge to gain competitive advantage. Elephant traps are waiting, though.
A common approach at the edge is to be hyper-contextualized to cut down the processing time, and therefore the overall latency. But serious dangers lie in automating decisions based on data concerning a single entity or a single set of events because inevitably they result in narrow decisions. And narrow decisions are not usually what deliver differentiation in business, nor return on investment (ROI). Differentiation is based on sets of complementary features that create capabilities rather than individual features, which as stand-alones do not set the pulse racing.
We see the obvious places for the edge as being at the perimeter of the mobile operator’s network or on the enterprise’s premises. Some applications, like industrial automation in a factory, have a pretty clear definition of what edges are, but this doesn’t apply in a massively distributed environment, let’s say like an oil field or huge construction site where safety is a serious concern, or a plane in flight or agriculture on the US’ Great Plains.
In agriculture, maximum food production might depend on ongoing, immediate minute adjustments of massive machinery taking into account the soil composition in each part of a field of several square miles, humidity, type of seed, weather forecast, depth of planting, an unforeseen obstacle and so much more.
Predictive maintenance and digital twins are growing into a sizeable market, but no matter how good the predictive algorithms, they will never be able to entirely replace the need to adjust or stop machinery, fast, when something entirely unforeseen goes wrong. It could be something like a lubricant has become more viscous due to a higher temperature in the machine causing coagulation. This might only be apparent and understood from assimilating a number of data sets.
Very low latency 5G networks are on the way, and incremental improvements are likely to go on indefinitely. Wise enterprises – and those who provide them with services – will not be focused on the speed of the network’s round trip, but how they can both gain competitive advantage from it.
So what do enterprises want to bring to the edge? The first thing to recognize is that the value is not in how few milliseconds (ms) the network can achieve the round journey, but in combining it with a complete package of capability into ecosystems that deliver maximum value from the speed at which complex decisions are automated.
The owner of the platform that delivers that capability holds the keys to the digital kingdom. Telcos spend a lot of time talking about moving up the value chain beyond connectivity, this is a great opportunity.
Where value lies
It’s often hard to put exact figures on the benefits of low latency, so this example made quite an impression. Mobile operator Comviva (formerly Mahindra Comviva), has an application called MobiLytix which pulls data from call detail records and other data about customers to present the best-option offer(s) for them in under 10ms. Does the 10ms matter?
It’s crucial: Although the total offer turnaround budget is 250ms, the offer needs to be presented within 4 to 7ms to have the greatest chance of succeeding; any longer and the propensity to accept offers falls steeply and most of that budget is taken up by the assembly of the artefacts needed to render the offer on screen. Getting the offer in front of the customer within 7ms was achieved by running VoltDB’s in-memory data technology which is designed to automate super-complex decisions at extremely low latency.
This resulted in a 253% improvement in offer acceptance for the operator and is a great demonstration of the hidden value in enabling sub-10ms decisions. When you start to think about it, it seems the rapid, data-driven decision capability in conjunction with super low latency networks could be applied all over the shop as enterprises eye the edge to gain competitive advantage. Elephant traps are waiting, though.
Value at the edge
A common approach at the edge is to be hyper-contextualized to cut down the processing time, and therefore the overall latency. But serious dangers lie in automating decisions based on data concerning a single entity or a single set of events because inevitably they result in narrow decisions. And narrow decisions are not usually what deliver differentiation in business, nor return on investment (ROI). Differentiation is based on sets of complementary features that create capabilities rather than individual features, which as stand-alones do not set the pulse racing.
Where is the edge?
We see the obvious places for the edge as being at the perimeter of the mobile operator’s network or on the enterprise’s premises. Some applications, like industrial automation in a factory, have a pretty clear definition of what edges are, but this doesn’t apply in a massively distributed environment, let’s say like an oil field or huge construction site where safety is a serious concern, or a plane in flight or agriculture on the US’ Great Plains.
It is well-known that air disasters often do not result from the original problem but from the sequence of events they set off, delays in understanding what’s going on and how one thing has impacted something else.
In agriculture, maximum food production might depend on ongoing, immediate minute adjustments of massive machinery taking into account the soil composition in each part of a field of several square miles, humidity, type of seed, weather forecast, depth of planting, an unforeseen obstacle and so much more.
Predictive maintenance and digital twins are growing into a sizeable market, but no matter how good the predictive algorithms, they will never be able to entirely replace the need to adjust or stop machinery, fast, when something entirely unforeseen goes wrong. It could be something like a lubricant has become more viscous due to a higher temperature in the machine causing coagulation. This might only be apparent and understood from assimilating a number of data sets.
Very low latency 5G networks are on the way, and incremental improvements are likely to go on indefinitely. Wise enterprises – and those who provide them with services – will not be focused on the speed of the network’s round trip, but how they can both gain competitive advantage from it.
