As manufacturers and other industrial organizations continue to leverage the Industrial Internet of Things (IIoT), we’re seeing a fundamental shift in industrial architectures. This isn’t about pilot projects either; we’re talking about large-scale production implementations. In fact, according to a recent survey of firms in 2022, nearly 62% of the companies surveyed have already deployed or are in the process of implementing an IIoT strategy (IIoT World, 2022).
What makes this paradigm shift different from others is the move from prioritizing automation hardware to a focus on data, as well as the software-defined applications, edge computing architectures, and cloud-based solutions that govern data’s movement in IIoT manufacturing systems. This new focus on data and software-based solutions is well underway. This year, for the first time, the average manufacturer will spend more on industrial software than on industrial automation hardware (OT hardware), according to IoT Analytics’ latest report on the industry titled Industrial Software Landscape 2022–2027 (see fig.1).
The benefits of this shift to virtualized infrastructure, in terms of business agility, improved production, and streamlined operations, have been well-described elsewhere. The key aspect of this new reality we wish to focus on for this post is the use of cloud-based platforms to manage movement of data through the IIoT deployment, particularly in terms of analytics, data storage, and more. The challenge is that as IIoT systems generate more and more data with ever-increasing velocity, how will manufacturers manage, leverage, and secure the flow of data to the cloud platform of its choice? Does that cloud platform have the capability, reliability, and performance to ingest data generated by today’s (and tomorrow’s) IIoT systems? The average factory generates terabytes of data daily and this volume is expected to grow rapidly (IBM, 2022).
With that in mind, manufacturers should consider the following criteria when they are evaluating various cloud services for ingesting IIoT data.
Compliance with Industry Standards
This is arguably the most important of the criteria organizations should consider. Does this service conform to published standards like MQTT and Sparkplug? There is a powerful consensus that these two standards are key to the future of the IIoT. Cloud services that support these standards should offer predictable performance and reliability that proprietary solutions simply cannot guarantee. What’s more, without standards compliance, interoperability with other ecosystem solutions is jeopardized, threatening the manufacturer’s ability to work with partners and severely limits its access to other technologies. Finally, solutions that do not adhere to industry standards could result in “vendor lock-in,” giving the cloud platform undue power over the strategic direction of the manufacturer’s IIoT future. Ostensibly, the move to cloud solutions was a means of avoiding this problem, so choosing services that do not conform to common standards doesn’t make a ton of sense.
Scalability
The one thing almost any manufacturer can guarantee is that the amount of data they will generate continues to grow. As a result, the ability of your cloud platform to scale to meet these ever-growing demands is paramount. Performance features to evaluate include the number of connections that a cloud platform can support, as well as the number of messages per second.
A cloud platform’s ability to scale message implementations is also key. A great example is MQTT topics, which are a form of addressing that allows MQTT clients to share information. MQTT Topics are structured in a hierarchy similar to folders and files in a file system. A cloud service’s ability to parse these messages at a performance level your application requires is important.
Reliability
A separate issue from performance, the ability to provide bullet-proof reliability in a manufacturing or industrial setting is a mission-critical element for any cloud-based ingest service. The data generated by IIoT systems does not stop and that means the cloud platform needs to deliver a matching set or reliability. This is historically a big issue for cloud-based services. It seems we hear of a cloud outage almost every week. As a result, it behooves industrial organizations to hold their cloud vendors to strict, well-defined policies and QoS levels, as well as implement contingency plans for when (not if), these cloud services experience an outage.
Observability
The ability for organizations to own and observe their own data is theoretically one of the primary reasons for moving to data-driven analytics solutions that are part and parcel with most IIoT deployments today. When evaluating cloud platforms that will connect to your IIoT network, you need to ask multiple questions about the ability to observe your data in action. Do your IoT devices turn into opaque “black boxes” when they connect to the IoT service of the cloud provider? Can you query each individual device to understand their status? These are questions you need to ask your cloud provider in order to determine if they’re worth their salt.
Observing their IoT/IIoT environment for discovering trends, anomalies and providing day-to-day visibility is among the primary reasons for moving your data to analytics solutions in the cloud. For example, if you are looking to make decisions based on real-time analytics, the analytics service for your cloud provider should give you (or your systems) visibility into that information. Alternatively, if a cloud provider offers to manage your IoT devices, you should ask if they can help you manage individual devices. These capabilities are important as many IoT services can easily turn into “black boxes” that don’t show you what’s happening with your data. Providers’ answers on these issues will help you determine the right provider for your Observability needs.
Flexibility
Rigidity in terms of deployment is never a good look for an IIoT-based cloud service. Most services today should be able to support hybrid models for both on-premise and cloud deployments for the same IIoT application. Their service should also be flexible enough to adjust to the unique needs of your IIoT deployment. Also all IIoT deployments have unique aspects native to their organization, so a good ingestion service for cloud data should be able to accommodate your network. Examples include things like device-onboarding and security policies. Be sure to quiz your platform vendor on these and anything else specific to your deployment.
Making a Final Decision
While there are undoubtedly other criteria that your organization may require in order to successfully navigate the myriad cloud-based data ingest solutions available today, these five criteria are still a powerful foundation upon which you can build your evaluation protocols. Key to your success in this arena is to also fully understand your own needs and what priorities you need to give to your specific mix of applications. As our industry continues its transition to a data-defined future, hopefully these guidelines help you to define your own future for data-driven IIoT architectures.
Leave a Reply