The Intelligent Edge Layer - what is it and why should you care?
One of the biggest challenges that Enterprises have in their efforts of digitalization is the fact that they have too many, and complex, data silos. Another is that they have a patch-work of applications that don't follow a common architecture. The promises of Industry 4.0 / Industrial IoT are significant - including big savings in operational cost, increased uptime of assets and processes, higher productivity and output, and increased competitive strengths. But there are a few big traps that companies need to avoid in order not to risk the long term outcome of the digitalization projects:
- Avoid building new data silos.
- Avoid cutting corners by solving use cases one by one with a different tech-stack every time.
- Vendor lock-ins from vertically integrated solutions will limit your flexibility.
- Don’t make choices that limit the ability to innovate in the future without costly redesign or new investments.
The answer? Focus on an architecture that is flexible, open and future-proof.
What is the The Intelligent Edge Layer?
With an Intelligent Edge Layer we mean a smart integration solution that you can use to build real-time integrations between your systems, whether it’s machines on the factory floor or enterprise systems that are used to operate your organization.
- We call it Intelligent because it’s not just about transferring data between two systems, but adding advanced logic to the integrations as well.
- We call it Edge because you run these integration solutions close to the systems you want to integrate, so that you can build integrations between systems that are only accessible locally, where you don’t want to expose data outside your security boundaries or where you want to reduce the data volume and only send relevant information over your networks.
- We call it Layer because it sits in between your existing systems and can integrate both horizontally and vertically across your existing infrastructure.
In the end of this article we will look at some of the Lessons Learned but let's first look at why it matters.
Why should you care?
Adding an Intelligent Edge Layer in your architecture gives you a number of advantages:
- An unlimited flexibility to add new use cases, at any time
- A platform that brings together OT, IT and subject matter experts
- A way to deploy one use case at the time but re-using the same architecture
- Great sleep at night as you now have a future proof architecture that will allow you to scale fast and cost-efficiently while leveraging your existing resources
- An increased life-length of your existing systems as intelligent logic and automation workflows can be done in the Edge Layer
Unlimited Flexibility and Future-Proof
An Intelligent Edge Layer is flexible and complements any existing architecture that you might have (i.e ISA-95) and enables a wide variety of use cases, such as:
- Machine-to-machine communication, e.g. for optimizations or anomaly detection.
- Advanced analytics, trigger actions based on analysis of your real-time data.
- AI & ML at the edge. Leverage custom algorithms to improve quality/yield, reduce raw material usage, reduce power consumption or to optimize your process.
- On-premise to Cloud integrations, integrate your on-premise systems with cloud systems without exposing the on-premise systems.
- Extend the existing technology stack. Add use cases not currently supported by the existing technology or simplify integrations by bypassing layers and build direct integrations between the relevant systems.
Leverage new data types, such as images, video and audio to optimize your processes and build new intelligent workflows.
Start Small. Or Go Big.
We have seen two distinct approaches when introducing an Intelligent Edge Layer. It all depends on the scope of the initiatives but no matter your initial approach there are some key considerations to bear in mind.
Being Tactical (start small)
Many of our customers have a pressing short-term need and start from there. Here you minimize the startup time by identifying some low-hanging fruit, i.e. some initial use cases which are clearly understood, reasonably simple and where you know you will get quick returns once they are up and running.
But the starting point doesn´t have to be a new use case. Sometimes there are technical challenges or problems on the shop-floor that are the starting point. Simply getting hold of the right data from OT or IT, transforming and integrating it to an on-premise or cloud base storage system could cause enough headaches.
The most important consideration is to avoid the traps discussed above. Don´t build a new silo or take corners when selecting the tech-stack. It might seem like a great way forward to select a vertically integrated solution that has data collection, logic, storage and visualization included. Until you are faced with your next use case and realize the solution you have selected isn’t flexible enough to scale with your needs and on top of that gives you all the disadvantages of vendor lock-ins.
Instead, take a longer view and consider how you can grow with a solution, indefinitely. Even if you start small there will always be new challenges and opportunities and if you take an architectural approach and implement an Intelligent Edge Layer you will have a flexible and future-proof solution that will scale with your future needs. Without a more costly starting point.
Being Strategic (start Big)
More and more customers are starting with a larger scope and here it is more obvious to start with defining an architecture and a common data model. A typical approach is to start with a few use cases but already from the start there is a roadmap of future use cases that is part of the scope. The Intelligent Edge Layer is the natural conclusion.
The upsides here are that you are more likely to come up with a long-term solution that doesn’t have to be revised. Another key value is that you will see exponential innovation pace. Even if the initial startup time is longer, clear architecture and data models will help you implement new use cases in a more efficient/simpler way, which over time will overtake the initial fast startup with the tactical approach.
Specific traps to this approach is to think that a message bus / integration layer without real-time analytics and advanced logic capabilities is enough or that all intelligence is to be performed in a cloud service.
Lessons learned
Finally we want to share some lessons learnt when working with customer projects.
Not one large project, but many mini-projects
No matter if you start small or large we advocate an approach where you shouldn’t see the actual use cases as one large project. There will be many use cases over time and you should start with the small and simple ones and gradually implement more complex use cases. See it as “eating the elephant in pieces”, otherwise it will be overwhelming and you may never get the expected benefits. This also means that you should plan for many small projects already from the start, i.e. consider things like structure and naming conventions.
The great thing is that an Intelligent Edge Layer can gradually be implemented in a flexible way as it complements your existing infrastructure and architecture. You build it out step-by-step as you add more use cases or sites. This means that the entry cost is small and scales as you grow and create value with your use cases.
Many integrations require logic and action
Simple use cases may be to just transfer data from one system to another, possibly with some transformation of data formats. However, great value can come from the use cases where you derive some insights from the data by applying logic to your use case. This could be simple condition logic all the way to applying advanced algorithms or ML models. With logic as part of your integrations you can trigger actions instead of just sending data, or you can enhance the value of the data by extracting features already next to the source and offloading the receiving systems.
Once you have an Intelligent Edge Layer in place you can start to build smart integrations without having to build logic in the endpoint systems. It’s enough to learn the integration layer and then just connect to other systems. This is a dramatic simplification compared with the traditional approach where the intelligence is built into each endpoint system, all with different technologies and tools. It can also extend the life length of your existing systems as you can enhance them with intelligence in the integration layer.
Separate data collection and harmonization from use cases that use the data
As you will see over time many use cases will need the same (or a subset of) data. Hence it’s advantageous to separate the data collection and harmonization from the actual use cases that will use the data. Then new use cases can be added without impacting existing use cases, even if the same data is used. If data is published based on a common data model it also means that new use cases can be built without knowing the details of each source and new data sources can be added without impacting existing use cases.
A message broker architecture is very useful to implement this type of setup.
Low-Code = No-Code plus Your Code.
Even though we at Crosser are strong advocates for low-code design, to enable as many as possible to be able to implement these types of integrations, you shouldn’t forget that we are still “developers” in some sense. This means that also with low-code implementations there are standard developer practices that are very relevant. Naming conventions, error handling and test procedures are as important also with low-code designs and will help you build stable and maintainable use cases.
No software solution has everything you need and sometimes using your own code or algorithms makes sense so having the flexibility to combine no-code with your code gives you a freedom to address every use case you want to implement.
Centralized management
The common raised objection to implementing software in the edge is the increased life cycle challenges and cost it brings combined with a lack of visibility. Modern solutions have addressed this with a centralized management and orchestration solution that automates and simplifies every aspect of the life cycle. Making it just as easy to operate edge software as a cloud environment.
To learn more about Crosser and our Edge Solutions book a demo with one of our experts.