In enterprise IT circles, it’s hard to have a conversation these days without talking about edge computing. And there’s a good reason for this. “The edge” is where businesses conduct their most critical business. It is where retailers transact with their customers. It is where manufacturers produce their products. It is where healthcare organizations care for their patients. The edge is where the digital world interfaces with the physical world – where business critical data is generated, captured, and, increasingly, is being processed and acted upon.
This isn’t just an anecdotal view. It’s a view backed up by industry research. For example, 451 Research forecasts that by 2024, 53% of machine- and device-generated data will initially be stored and processed at edge locations. IDC estimates that, by 2024, edge spending will have grown at a rate seven times greater than the growth in spending on core data center infrastructure. In a word, this kind of growth is enormous.
What’s behind the rush to the edge? The simplest answer to that question is that business and IT leaders are looking for every opportunity they can find to achieve a competitive advantage. Eliminating the distance between IT resources and the edge achieves several different things:
- Reduced latency– Many business processes demand near real-time insight and control. While modern networking techniques have helped to reduce the latency introduced by network hops, crossing the network boundaries between edge endpoints and centralized data center environments does have some latency cost. You also can’t cheat the speed of light, and many applications cannot tolerate the latency introduced by the physical distance between edge endpoints and centralized IT.
- Bandwidth conservation– Edge locations often have limited WAN bandwidth, or that bandwidth is expensive to acquire. Processing data locally can help manage the cost of an edge location while still extracting the maximum business value from the data.
- Operational technology (OT) connectivity– Some industries have unique OT connectivity technologies that require specialized compute devices and networking in order to acquire data and pass control information. Manufacturing environments, for example, often leverage technologies such as MODBUS or PROFINET to connect their machinery and control systems to edge compute resources through gateway devices.
- Business process availability– Business critical processes taking place in an edge location must continue uninterrupted – even in the face of a network outage. Edge computing is the only way to ensure a factory, warehouse, retail location, or hospital can operate continuously and safely even when it is disconnected from the WAN.
- Data sovereignty– Some industries and localities restrict which data can be moved to a central location for processing. In these situations, edge computing is the only solution for processing and leveraging the data produced in the edge location.
As companies implement edge computing, they are moving IT resources into OT environments, which are quite different from the IT environments that have historically housed enterprise data. IT teams must adapt IT resources and processes for these new environments.
Let’s talk about the state of many edge implementations today and how to optimize your path forward.
Moving Beyond Proofs of Concept (POCs)
The process of implementing and operating edge computing isn’t always straightforward. Among other things, edge initiatives often have unclear objectives, involve new technologies, and uncover conflicting processes between IT and OT. These challenges can lead to projects that fail to move from the proof-of-concept stage to a scalable production deployment.
To help organizations address these IT-OT challenges, the edge team at Dell Technologies has developed best practices focused on moving edge projects from POCs to successful production environments. These best practices are derived from our experience enabling IT transformation within data center environments, but they are adapted to the unique needs of the edge OT environments. To make this easy, we have distilled these best practices down to three straightforward recommendations for implementing edge use cases that can scale and grow with your business.
- Design for business outcomes.
Successful edge projects begin with a focus on the ultimate prize — the business outcomes. To that end, it’s important to clearly articulate your targeted business objectives upfront, well before you start talking about technology. If you‘re in manufacturing, for example, you might ask if you want to improve your production yields or to reduce costs by a certain amount by proactively preventing machine failure and the associated downtime.
Measuring results can be difficult when you are leveraging a shared infrastructure, especially when you are trying to look at the return on investment. If your project is going to require a big upfront investment with an initial limited return, you should document those business considerations and communicate them clearly. Having specific business goals will enable you to manage expectations, measure your results as you go, and make any necessary mid-course corrections.
- Consolidate and integrate.
Our second recommendation is to look for opportunities to consolidate your edge, with an eye toward eliminating stove-piped applications. Consolidating your applications onto a single infrastructure can help your organization realize significant savings on your edge computing initiatives. Think of your edge not as a collection of disconnected devices and applications, but as an overall system. Virtualization, containerized applications, and software-defined infrastructure will be key building blocks for a system that can enable consolidation.
Besides being more efficient, edge consolidation also gives you greater flexibility. You can more easily reallocate resources or shift workloads depending on where they are going to run the best and where they are going to achieve your business needs. Consolidating your edge also opens opportunities to share and integrate data across different data streams and applications. When you do this, you are moving toward the point of having a common data plane for your edge applications. This will enable new applications to easily take advantage of the existing edge data without having to build new data integration logic.
As you consolidate, you should ensure that your edge approach leverages open application programming interfaces, standards, and technologies that don’t lock you into a single ecosystem or cloud framework. An open environment gives you the flexibility to implement new use cases and new applications, and to integrate new ecosystems as your business demands change.
- Plan for growth and agility.
Throughout your project, all stakeholders must take the long view. Plan for your initial business outcomes, but also look ahead and plan for growth and future agility.
From a growth perspective, think about the new capabilities you might need, and not just the additional capacity you are going to need. Think about new use cases you might want to implement. For example, are you doing some simple process control and monitoring today that you may want to use deep learning for in the future? If so, make sure that your edge infrastructure can be expanded to include the networking capacity, storage, and accelerated compute necessary be able to do model training at the edge.
You also must look at your edge IT processes. How are your processes going to scale over time? How are you going to become more efficient? And how will you manage your applications? On this front, it makes sense to look at the DevOps processes and tools that you have on the IT side and think about how those are going to translate to your edge applications. Can you leverage your existing DevOps processes and tools for your off-the-shelf and custom edge applications in your OT environment, or will you need to adapt and integrate them with the processes and tools that exist in your OT environment?
A few parting thoughts
To wrap things up, I’d like to share a few higher-level points to consider as you plan your edge implementations.
Right out of the gate, remember that success at the edge depends heavily on having strong collaboration between your IT stakeholders and your OT stakeholders. Without that working relationship, your innovations will be stuck at the proof-of-concept stage, unable to scale to production, across processes, and across factories.
Second, make sure you leverage your key vendor relationships, and use all the capabilities they can bring to bear. For example, Dell Technologies can help your organization bring different stakeholders within the ecosystem together through the strong partnerships and the solutions that we provide. We can even customize our products for particular applications. Talk to us about our OEM capabilities if you have unique needs for large edge applications.
Finally, think strategically about the transformative power of edge, and how it can give you a clear competitive advantage in your industry. But always remember that you are not the only one thinking about edge. Your competitors are as well. So don’t wait to begin your journey.