Dream big, start small – secrets of IoT project success from an Azure IoT expertAdvice
The Device Chronicle interviewed Stefanie Grois, IoT Solution Architect, Microsoft Azure IoT. Stefanie shares some vital wisdom and important considerations from the front line of implementing IoT projects.
Stefanie begins by identifying the three most common considerations when setting up an IoT project that will go from proof of concept to production. The first one is for project leaders to dream big but really start small. What this means is they must not start by addressing the technical issues, connecting devices and so on. Rather, it means to focus on the business motivation for the project, how it will help the business through increasing efficiency, saving costs and being able to articulate this powerfully to various key stakeholders across the organisation.
Start with the overall vision
For Stefanie, it is vitally important to shape an overall vision for the project early and one that resonates with both the technical teams and business decision makers. What direction of travel should this IOT project take and what is its major goal? In Stefanie’s experience, the most successful projects involve multiple departments working together cohesively with decision makers and management. “Dream big, start small actually means that you should have this vision and a business model to support it.”
Start small and evolve
The second consideration for Stefanie is to focus on a specific use case to prove the IoT approach and then compound by building on top of this initial success. Stefanie believes that this is a horizontal topic that traverses so many different technology areas, processes and departments in an organisation, that it is more prudent to start out with a very distinct project with clearly demarcated boundaries. It is important to select a use case which will deliver immediate value out of it or solve a serious pain for the organisation. It can be really easy to remove this pain leveraging cloud and IoT technologies when compared to what the organisation was doing beforehand. Stefanie explains “A simple example could be a production machine on the shop floor that has been continuously failing and where no one has a clear reason as to why this is happening as there is no data coming out of the machine to analyse and understand the fault. The situation changes once an analysis of the machine data reveals the underlying reason for the fault.”
Control the project holistically
The third consideration for Stefanie is whether the IoT project leader actually has ownership of all the processes and mechanisms that will enable a successful outcome for it. Many IoT project leaders can overthink the project and make it very complex. They can have an overall strategy and vision, but then they get so afraid of the complexity of the project that they have just envisioned that they cannot find a way to start small. Even if they have a use case, they ask if this use case is dependent on another one and if so, what will they do then? It leads to procrastination and project paralysis from fear of failure.
So IoT project leaders must in Stefanie’s estimation be agile and avoid a waterfall model where they go from one step to another envisioning, and then implementing the solution and then thinking they are completed in one linear direction. They must use an agile model where they always go through the same circle again and try to improve.
Stefanie has also experienced situations where the organisation believes that they have already solved the use case on premise but that actually multiple data sources and insights are missing to allow for the overall vision to be realised. They believe that they have all the insights they need, but then it appears that there is a massive information gap between the shop floor department and the business decision makers. And this speaks to how organisations can be isolated in information silos so you have situations where one department just decides to start an IoT project but they do not have the ownership or the influence to drive the project in one direction for the strategic benefit of the organisation. This is where having a trained management consultant on the IoT project can help ensure that a more holistic approach is taken to consider who exactly from across the organisation should be involved in the initial consultations on the IoT project.
Prepare to scale from the outset
Stefanie also advises IoT project leaders to think about scale from the very outset of the project. They must be able to move from proof of concept to production (and the scale that comes with this) in a frictionless manner. In many IoT projects, the project leaders tend to make a lot of compromises just to get the use case working. And then if they need to scale out because they are releasing more devices, they struggle because they have built their entire architecture on a very small basis and very distinctively for one use case. So the system architecture is not modular and may not have any automation integrated into it. So there is no easy way to deploy tens of thousands of devices, all processes have to be triggered manually by employees. Stefanie believes these organisations also make the mistake of conflating the need to scale with developing their own homegrown platforms but leads to increasing complexity and heavy investments of time and money. IoT project leaders should look towards best of breed managed point solutions to accelerate their time to launch and ensure they get the automation features, robustness and security they need to scale in line with their projections.
Think end to end
Stefanie also advises IoT project leaders and their organizations to think end to end. Beyond the technology implementation, they must think about the actions for the business, how the peer systems will trigger an action, if the IoT use case is predictive maintenance then how from the AI trained model will the relevant support teams be triggered to complete an action through the system?
Build a budget for continued development
To be successful in an IoT project, Stefanie says of course the technical requirements and business requirements must be fulfilled, but a budget for sustainable development and innovation must also be established. This is in line with agile thinking and not falling into the trap of thinking that the project will have a finite completion date in time. The IoT project leaders must be able to drive the project over several years.
Performing knowledge transfer with a partner who has already set standards from multiple projects is also a good idea. Equipped with greater experience, this partner can make them aware of things that can help them avoid certain pitfalls in a project.
Example IoT use cases
Stefanie also says that the most impactful projects are the ones that are able to bring the heterogeneous data sources into a single control panel and then drive actions from here. Then there is a whole data platform that brings everything together. This is really what an IoT project is about. It was once thought to be impossible to get insights from containers on freight ships all over the world. To answer questions instantly such as what is the temperature in a certain container? Or is there a failure in the cooling system in a certain container? Has a certain container been stolen by pirates or compromised in some other way? The Azure team solved this challenge with shipping and logistics leader Maersk. Such IoT innovation helps the solution provider and the customer to understand the root causes of problems and identify challenges in service provision. Stefanie says that armed with the data insights from connected assets, it is easier to understand “Who has the liability in a certain situation? Is what is being claimed really true? What is the exact location of the container? Decisions can be made based on facts.
Another powerful use case is pumps on an oil rig. Stefanie says it is necessary to monitor machines on oil rigs to see if they are failing and if a service engineer needs to be dispatched to fix it. She says “This engineer is not necessarily living on the oil rig so you can imagine how expensive it is to send someone there. But armed with data insight from the performance of the pumps, you may be able to better control when you have to dispatch a service engineer.”
In a power grid management project, Stefanie explains that Azure partner Robotron is doing real-time measurement of relevant energetic quantities in (low)-voltage nets with non-regulated measurement technology. Through the power of the cloud, the service provider in this case can get vital information about the Power Grid in real-time and analyze historical data to find patterns and anomalies to keep the power grid operational and balanced.
Stefanie provides another EV Charging use case example: Microsoft Azure partnered with NXP to create an infrastructural solution to monitor power consumption at peak and off peak time through the EV charging units with charging data being transferred directly from the processors in the units to the cloud for analysis. It provides a single control plane to gain insights on EV power consumption so the City can better understand its energy capacity requirements. The goal of this EV charging case is to get a better grasp of the usage patterns of the EV-Charging units, for example which unit is often used and when, what is the customer profile, what kind of cars are charging there and so forth and get the information directly from the Smart chip in the charging unit. This information is of great use for Grid Operators (as explained in the use case above), Charge Point Operators, OEMs and customers because they can see their personalised and important information that has been collected via Azure RTOS and visible to them with the analytics capabilities in the cloud.
An Azure expert
Stefanie and her team work with a wide range of organizations across different industry sectors to advise and implement IoT projects leveraging the Azure, Azure IoT and Azure Edge IoT infrastructures and toolsets to meet their project requirements and achieve business goals.
Mender has a reference integration with Microsoft Azure IoT for robust and secure OTA software updating of embedded devices in a fleet at scale.
Microsoft offers a powerful set of SDKs for IoT devices. Stefanie says that the SDKs are very important because they are actually the clients that connect, and they have the built-in libraries to connect to Azure IoT very easily. Stefanie believes that it is important to automate as many things as you can through the available SDKs and APIs. Azure IoT Hub APIs, for instance, can be integrated with Azure DevOps, Github Actions or something similar to automate the whole pipeline or to retrain, for example, machine learning models designed to work on the IoT devices.
For the shop floor and the manufacturing customers, Stefanie says IoT Edge is critical. Azure offers runtime services for edge devices so this could be industrial PCs that are running on the shop floor and can run independently of the cloud and disconnect from the cloud, but can also be provisioned through the cloud and if the management must be done through the cloud.
We wish Stefanie and her colleagues at Azure well as they continue their journey to help organisations successfully utilise IoT for business innovation and transformation.