With Internet of Things (IoT) networks continuing to add more devices, we explore how to go about developing an effective scaling strategy for IoT.
Scaling IoT networks effectively can be easier said than done. While having more devices leads to more connectivity, there is the danger of encountering issues with latency, as well as security. In addition, there is the service that’s being delivered to customers to consider.
“From experience, as true scaling starts – typically at around 1,000 devices deployed – providers hit a brick wall as they suddenly realise that they have neither the tools nor the processes to answer even basic questions about their device estate and customers, such as ‘Am I actually providing a good service?’”, said Pilgrim Beart, CEO of DevicePilot.
“More often than not the answer is “I don’t know,” as the quality of ongoing service delivery has been overlooked in favour of product development, which creates massive problems when scaling.
“Once the product is deemed to be ready, service providers start to scale and make promises to customers, which may even be contractual. If successful, they suddenly have many thousands of devices in the wild, collecting data.
“The spreadsheets and brainpower that had been used to monitor the devices become completely overwhelmed, and the provider has no idea what is happening on their devices – what is the device uptime and usage? How many are faulty? Are they capturing the right data?
“Without effectively monitoring their devices, providers can’t answer these questions. This is rarely front of mind when developing IoT products, but it makes scaling a nightmare.”
So, how can organisations ensure that their scaling strategy for IoT is the best it can be? Let’s find out.
Consider vectors and workloads
According to Martin Garner, COO at CCS Insight, companies have three aspects that they must take into account.
“There are three vectors for scale: More IoT projects across different use cases, using common infrastructure; scaling up each project from pilot to full deployment, and extracting more value per project,” said Garner.
“You need to plan for all of them. Spend plenty of time to get the architecture of the system right to enable these.
“You should also think carefully about what workloads should run where. IoT systems are typically distributed computing systems, and there are lots of options for optimising capital and operating expenditure.”
Determine business goals
As with any organisational strategy, scaling IoT needs clear goals from the outset for delivering value, as more eyes turn their attention towards the technology.
“When implementing IoT technology, there are very low barriers to entry – low cost, low technical complexity – so companies can have a tendency to dive right in without defining specific, measurable business goals,” said Brendan Mislin, industry x.0 IoT lead at Accenture. “This can lead to projects that are technically successful but don’t deliver significant or measurable values, meaning they quickly peter out rather than forming the foundation for a long-term, scalable strategy.
“In order to develop a successful, scalable IoT strategy, the business focus needs to shift from proof of concept to establishing proof of value. This requires selecting an advanced IoT platform that is stable and secure, to ensure the technology not just works but delivers true business value.
“Businesses need to evaluate whether an IoT use case can demonstrably save costs or increase revenue. Doing a short test of a particular use case over several weeks before full deployment is a useful strategy to test these values, helping to mitigate encountering issues down the line.
“As we enter an increasingly challenging global economic environment, scrutiny of IoT projects will only intensify – being able to showcase the proof of value in these projects will be key to long-term success and the ability to implement a strategy that is scalable.”
Consistent data management
An increase in devices in use across the organisation brings a need to effectively manage the data generated. This requires constant monitoring to ensure that the network is operating to the expected standards.
“IoT projects are growing, and companies are changing their approach based on having more data from more devices,” said Joshua Norrid, senior technical director at DataStax. “However, scaling up does test your assumptions that you have put in place around infrastructure and data.
“For example, are your devices creating data at the rate that you expected, or is there more data coming through? Are you able to manage that data at the edge or are you centralising it?
“For enterprises, this edge or centre model is not going to be flexible enough. You will want a consistent approach to data being stored and managed across edge, centre and in-between.
“If it is hard to use your data, then it will make it harder for your developers to build the applications that rely on that data over time. It can make it more difficult to create those innovative projects that use IoT data, which can affect the success of your plans around IoT.”
Staying on the topic of monitoring the network, the rise in IoT devices can mean that the technology is deployed in hard-to-reach areas, meaning that it can be more difficult to recover from outages.
Alan Stewart-Brown, vice-president EMEA at Opengear, suggests implementing a remote management plan in order to overcome this obstacle.
“One of the most often overlooked or under budgeted issues of IoT scaling is not the initial build out of the system which is typically well planned for, but the long-term maintenance and support of what can quickly become a huge network of devices that are often deployed in difficult to reach locations,” he said.
“That complexity requires a resilient network to ensure that all of these IoT devices, connected via an aggregation point, can be securely managed and updated to extend their lifespan. Where edge compute is necessary due to the density of connected IoT devices, it is also advisable to provide scalable, secure and highly reliable remote management for all the IoT network infrastructure that provides a fast and predictable way to recover from failures.
“An independent management network should provide a secure alternate access path, including the ability to quickly re-deploy any software and or configs automatically onto connected equipment if they need to be re-built, ideally without having to send an engineer to site. In general networking terms, it is very important to ensure that the IoT gateways and edge compute equipment stack is actively monitored and that it is designed with resiliency in mind.”
Lastly, Sam Lakkundi, vice-president, innovation at BMC, cites AI as a possibly key technology in helping to manage data while scaling IoT.
“Businesses should consider the role of edge computing in supporting the deployment of IoT,” said Lakkundi. “By moving certain workloads to the edge of the network, devices spend less time on the cloud, react faster to local changes, and operate reliably–even in extended offline periods.
“Notably, it enables companies to leverage data generated by IoT devices to make ops more efficient, improve customer experience, generate new revenue streams, and become autonomous digital enterprises.
“Underpinning all of this lies the processing, understanding, and reactions to this data. When looking to scale, feeding this data through machine learning or AI systems is a must. These may sit in a separate data centre or cloud provider or within a core mainframe.
“There are also various functionality questions that must be asked, such as ‘Are you using your IoT only to monitor?’, or ‘Are you looking to have your operations autonomously react and remediate with IoT sensor data?’, or ‘Is this for proactive, predictive maintenance?’. Finally, this moves into the realm of AIOps, which is a core tenant of any scalable IoT data strategy.”