Dr John Bates, CEO of testing automation company TestPlant, spoke at the recent Nutanix Next 2017 conference about the end-to-end realities of putting IoT systems through their paces.
With enterprise cloud company Nutanix now embracing the worlds of edge computing, IoT and intelligent network automation, it makes sense that it invited Thingalytics author Dr John Bates to speak speaking at its recent customer, partner and developer event in Washington DC.
Formerly of Apama, Progress Software, Software AG and SAP Plat.One, Bates says he turned down job offers from much bigger tech companies to take up the leadership role at the rather less well-known TestPlant. His appointment was announced in February this year.
TestPlant specialises in a range of tools that provide automated testing of IoT environments: functional testing, performance testing, load testing and also higher-level network emulation.
Read more: RealVNC engages virtual automotive testing software
How do we test the IoT?
Published in 2015, Thingalytics: Smart Big Data Analytics for the Internet of Things (to give the book its full title) provides readers with genuinely useful insights into how the combination of real-time analytics and smart algorithms can turn IoT-generated big data into “gold nuggets” of business insight.
And that vision, of the combined power of IoT and big data, feeds directly into Bates’s leadership of TestPlant. So how do the key profit principles of this ethos break down and how do the operational models that derive from them now shape the testing approach?
To answer these questions, we have paraphrased excerpts from Bates’s presentation at the Nutanix event. While not always exact quotes, we believe they accurately convey his meaning and message.
Thingalytics profit principle #1
We need to collect IoT data and ‘upsell’ it. Take tires (tyres) as an example: they wear out and we replace them. But what if a tire became a connected ‘thing’, with networked sensors to gauge wear & tear and then talk to a telematics box that essentially exists as an edge device and add in data regarding the location of the vehicle and so on.
What could we do with that information? Could this lead to a new business model? Instead of buying tires, for example, could this now lead us to being able to lease tires? Could it also allow us to sense where tires are under-inflated and then calculate how much extra petrol the vehicle is using per mile, because we now have this level of data analytics? There’s a lot of extra control here if we want it.
Thingalytics profit principle #2
The Uberization of everything is coming. We can now start to put smart lighting in cities that illuminate only when there are vehicles on the road. This leads us on to areas including predictive maintenance so that we know where and when lighting will need to be replaced.
A city management organization in this scenario can then more dynamically interact with service companies [and other related firms in the total now IoT-connected supply chain], who can tender for the contract to replace bulbs or provide other services more profitably.
Thingalytics profit principle #3
This principle concerns datacenter monitoring: we can now start using IoT building management systems to more effectively monitor how much heat each rack in a datacenter is pushing out. Simple Network Management Protocol (SNMP) has been used in the past to help manage and optimize sensors and edge devices.
Smart infrared cameras can be placed around the datacenter and a front-end analytics system can help look for hot spots of heat and so the whole system basically uses cheap sensors to do what other firms are spending millions of dollars doing.
Read more: IBM launches new security services for connected cars and IoT
Four principles to power IoT
Bates then moved on from his delineation of where the profit opportunity exists in the IoT to more specifically look at some of the factors that will power and drive the next stage of IoT evolution.
In terms of how we drive and power the IoT of tomorrow, let’s say that it breaks down into four major component parts. There is now a need to:
- Bridge across legacy Machine-to-Machine (M2M) protocols and new IoT protocols;
- Future-proof applications and so model around IoT ‘objects’, not devices;
- Build adaptive platforms with smart logic placement across devices, edge and cloud environments (a microservices based architecture helps to be able to place elements around just where you need to);
- Move to rapid plug-and-play IoT authoring.
What the profit realities and power mechanics lead us to is a new understanding of how we test the IoT that we are now building.
“The IoT is really breaking the model for how we think about application testing. The traditional monolithic model of yesteryear has moved onward through packaged applications to web and then onto to mobile,” said Bates.
“Now though, in the world of the IoT, we see that everything has a mobile front end with cloud at the back-end… This all leads us to the new digital enterprise (with its need for so-called digital user experiences) and so testing has to change.”
“The way applications are measured has to now be based around a measure of ‘customer delight’ (and quality and correctness) and Key Performance Indicators (KPIs)… This means that we actually have to change the way application code is tested… and this is tough, in many senses, as we connect to a lot of code that we don’t even own, because maybe we connect to it through Application Programming Interfaces (APIs) and through other connections.”
Bates insists that traditional object-based testing does not work. We now need to go out and really ‘drive’ those APIs and make sure that the correct output is brought in. IoT system testing means intelligently automating paths through the mobile front end and APIs and indeed all the way through the complete data path.
“The only way to test and measure the digital experience is through a ‘Digital Automation Intelligence’ engine, interacting with the app as a user would and, at the same time, testing remote APIs to ensure the distributed plumbing works as it should. It’s no longer about just debugging to the code — it’s about using and analyzing the running application using AI techniques,” said Bates.
Read more: Stress testing the IoT bedsprings
Takeaway-alytics
It’s no secret that the IoT had its power switched on well before we all new how bright the lights were going to be. If we go along with Dr Bates’ arguments, we can see, deep inside the software architecture models now being developed to drive the IoT, that there are data and application component connectivity points there which are not fit for purpose.
Yes, Bates runs a testing firm dedicated to IoT component mechanics, so he would say these things wouldn’t he? But deliberate naysaying aside, there could be lessons here in terms of the back-to-basics responsibility that developers now have if they want the IoT to work the way in should in the future.
Spinalytics? Nah… this stuff is real.