Bridging the gap between IoT and the enterprise

We are seeing a lot of interesting work developing in the industrial internet of things. IIoT is making manufacturing, transportation, healthcare and many other verticals smarter. This is self-evident at this point, as we see new use cases pop up every day. While this is hugely exciting, one area that seems to be lacking in the IoT movement is strong champions within traditional enterprise roles. The support of influential decision-makers will be required to drive transformation when adoption efforts are more significant.

When cloud and SaaS first began to take hold, it was the stakeholders in CFO and VP of sales roles that really drove a lot of that adoption. For example, when Salesforce.com would target accounts, it was targeting the sales team and attempting to avoid IT altogether. This was a major part of its success because ultimately the value of any enterprise technology comes in the form of either driving revenue or reducing cost.

Now, it is probably not a wise practice in IIoT, which is far more complex and has far wider consequences than CRM, to exclude IT from the conversation, but it does give insight into how to drive major enterprise adoption of IoT.

Traditional enterprise roles need to be part of the conversation. Currently though, network engineers, electrical engineers and, in some cases, software developers are the only major stakeholders in IoT conversations. These folks all have incredible ideas about how to transform business with IIoT. Many of them are pioneers and thought leaders in the space. Despite this, IIoT will not see the explosion that SaaS and cloud saw until stakeholders outside of engineering are involved in the conversation.

In some verticals, like manufacturing and supply chain, this is already happening. This is primarily because the stakeholders are tightly integrated into the process and typically have engineering backgrounds. Prior to IIoT, the data that stakeholders needed to drive value was already tightly integrated into the process. Since the Industrial Revolution, these stakeholders have been looking at how to drive less waste in manufacturing and other similar use cases. They already had the models and systems in place to evaluate this data — IIoT just made it faster and more efficient.

In other verticals, the gap between those leading the IIoT charge and those driving business value is significantly greater. In these verticals, the primary stakeholders are using CRM and ERP software to drive business value. The gap between those traditional enterprise software systems and IIoT is vast. It is difficult to include them in the conversation because it is difficult to get them the data. Often when they do get the data it has been summarized, sanitized or preprocessed by a data scientist outside their business unit. For these business stakeholders need to sink their teeth into IIoT — they need to be getting the meat and right now, they are getting the morsels.

A major challenge today is end users are treated as consumers of IIoT data, but instead need to be treated as participants in the conversation. ERP and CRM systems are not designed with IoT sensor data in mind. Business end users are not going to move off these systems. Data management systems designed for IIoT commonly take 24 hours or more to get the summarized, sanitized data to these systems. These conversations need to be bidirectional in nature, allowing business end users to make decisions that can be pushed back to the edge. The vast majority of IoT data management systems have been using time-series or streaming data technologies which do not allow for a bidirectional conversation, but rather provide a one-way flow of information. This is hugely problematic.

Imagine that you work in finance for the oil and gas industry and you notice trends from production data that could negatively impact price. This data is now 24 hours old. Your response also needs to be manual or in a separate system. Imagine that eventually the person spotting these trends is instead AI. It should be able to respond to this production data in real time and take real-time action to avoid waste.

In order to achieve the true promise of IIoT, where IoT sensor data is driving real-time business action and reaction to reduce cost and drive revenue, architectures need to accommodate bidirectional real-time flows of information. This will allow business end users to join the conversation and have a direct impact which will in turn drive interest, reduce cost and drive revenue and massive adoption.

In other words, the role of business in the IoT data value chain needs to change from a consumer to a participant. We can again look at the success of Salesforce.com for clues. When CRM was born, it was typically implemented by IT and the business end users were then trained on how to use the system. This resulted in business processes being dictated by IT to the business in many cases. Additionally, it was a model where CRM end users were simply consumers of their system. This allowed Salesforce.com business end users to be able to customize their own workflow, drive their own business processes and create a bidirectional flow of information. This resulted in an explosion of SaaS software offering the same capability and highly engaged end users.

IoT could follow the same model by implementing bidirectional data management practices. This is not complicated, and technologies exist today to make this possible. The problem is that most big data architectures today are mono-directional. It was not necessarily the intention to make them mono-directional, but this happened naturally as they evolved and became more complex.

The reason is that most big data architectures started out with a NoSQL database for scale and flexibility, but quickly realized that the analytical capability they needed for decision-making needed to be found elsewhere. Typically, an RDBMS was bolted onto this system, which had the analytical capability needed, however it lacked the scale and flexibility. Afterwards, data lakes were added to the mix and this is where things became very mono-directional. Data lakes are great at getting a broad picture of an enterprises data, however it is difficult to respond from data lakes and they are slow. Furthermore, once you have more than two data sources in the data value chain, it is difficult for the architecture to be bidirectional in nature. The outcome can be data easily getting out of sync — if you make changes from one system to another, the third system can often be left out.

As a result, in order to achieve a bidirectional data conversation with the business and the edge, new architectural patterns need to be adopted for IoT. This is why we have seen the growth in hybrid transactional/analytical processing, or HTAP, systems over the last few years. These systems are designed to facilitate a bidirectional conversation with the ultimate goal of making data-informed business decisions in real time. They have the scale and flexibility needed for IoT in their transactional capability, combined with the decision-making capability found in their analytical processing.

Companies interested in building successful and future-proof IIoT architectures should look to HTAP systems. This will allow them to quickly and easily include business end users in the conversation, which will in turn ensure that their projects drive true business value.

All IoT Agenda network contributors are responsible for the content and accuracy of their posts. Opinions are of the writers and do not necessarily convey the thoughts of IoT Agenda.