Edge computing goes hand-in-hand with several other prominent technologies, especially hybrid cloud and 5G. It’s also ideally suited for Internet of Things (IoT) devices and applications. Actually, edge and IoT are more than just good partners, so to speak: They are likely to increasingly depend on each other.
“Edge computing is what keeps our heads above the water in the massive flood of data streaming to and from IoT devices, where every millisecond counts – especially for use cases like healthcare monitoring and safety apps,” says Stephen Blum, CTO at PubNub.
[ Get a shareable primer: How to explain edge computing in plain English. ]
Edge and IoT: How it works
Edge does this by bringing data processing and other computing needs as close to the sensor or other device as possible, which reduces latency, among other potential benefits.
“Edge computing reduces the cascade of potential bandwidth bottlenecks and processes the data that matters, keeping it close to the source.”
“Rather than sending data to be processed on external [cloud] servers or at central data centers, costing precious seconds and additional resources, the computation takes place on the device or in the network itself,” Blum explains. “From there, the processed data can be delivered to its destination sooner. Edge computing reduces the cascade of potential bandwidth bottlenecks and processes the data that matters, keeping it close to the source.”
In fact, this is currently the most common architectural pattern, according to Saurabh Mishra, senior manager of IoT at SAS: an edge computing environment placed near the sensors – or the “things” in IoT – that generate data.
“IoT and edge are intricately connected,” Mishra says. “By definition, IoT is about ‘things’ – things that are distributed and things that are connected with each other or a centralized infrastructure (like cloud) through a compute environment in the field.”
These compute environments might take a variety of forms, Mishra explains, from a remote server (aka edge server), a gateway, a switch housed inside a cell tower, a retail store’s back-office infrastructure, or a connected car.
[ New to edge? Check out our primer: How edge servers work. ]
“These compute environments are what support edge computing because they are small compute units distributed away from the core (like cloud) and have the capacity to perform a variety of tasks,” Mishra says.
Edge and IoT: Cost and flexibility benefits
Much of this data may be ephemeral, and a round trip to the cloud may not actually create any value.
Beyond performance and latency advantages, Mishra says, this can also be the most economical architectural choice. Among other reasons: Much of this data may be ephemeral, and a round trip to the cloud may not actually create any value.
There are other potential cost optimizations, such as reduced cloud spend or data center footprint, as well as advantages in areas like security.
“In executing business logic on the devices as close to the edge as possible, it reduces traffic sent to external servers and makes it so you don’t have to continuously add data center capacity to deal with growth,” Blum says. “That means better performance (no need to wait for sending and receiving data), lower operational costs, and higher security [via] limiting outward connections.”
The overarching appeal of this model – and a reason it also fits well with hybrid cloud architectures – is that it’s essentially a “have your cake and eat it too” deal.
“The big benefit of this new model comes down to allowing enterprises to have the best of both worlds: being able to sense, capture, and analyze massive amounts of data at the location of creation and obtaining global visibility, management, and deeper analysis – or even the creation of machine-learning models – in the cloud,” says Eva Schönleitner, CEO at Crate.io. “This edge model is one of the most critical enablers for successful digitization initiatives across IoT and industrial IoT use cases, such as with smart factories or smart buildings that are pumping out massive amounts of sensor data continuously.”
Organizations using a hybrid cloud strategy and edge computing in tandem gain flexibility and consistency.
“Businesses need flexibility in terms of where they place their workloads, and if their strategy changes, [they need] consistency of operations – for both ITOps and developers, so as to enable them to react quickly and with minimal disruption,” Rosa Guntrip, senior principal marketing manager, cloud platforms at Red Hat, recently noted.
[ Want to learn more about implementing edge computing? Read the blog: How to implement edge infrastructure in a maintainable and scalable way. ]
Examples where edge makes sense
Mishra from SAS notes that a variety of other trends, environmental characteristics, and business requirements can further drive more specific or more specific architectural decisions and use cases. Here are a few examples:
- Is the edge environment connected to a centralized hub, like a cloud or a traditional data center? “A dedicated on-premises environment like a physical retail store or a manufacturing facility will generally have a dedicated connection to a central hub [such as a cloud],” Mishra says. A moving locomotive or offshore oil rig, on the other hand, may have more sporadic connectivity: “These connectivity considerations dictate the use cases that can be supported at the edge.”
- Do you need to perform control logic locally? This requirement matters a great deal in some scenarios – Mishra says it’s a must-have for autonomous vehicles, for instance – and not so much in other contexts. “In a factory that has a large number of machines and processes, it’s often a goal to better use the data generated to drive the local control logic,” Mishra says. “At the other extreme, in an industry like healthcare, there is not an immediate need to perform local control logic and the use case supported by edge computing could [instead] be store-and-forward or asset tracking.”
- Have standards been established and adopted? Today’s reality is somewhere between “sort of” and “not yet.” This is one reason behind the growing interest in multi-access edge computing (MEC), which is both a way of thinking about the outermost edge (especially in a 5G world) and an actual standards framework. “An MEC pattern promises to open new edge ecosystems and value chains with support for horizontal use cases that range from video analytics to location-based services to augmented reality,” Mishra says.
This latter point speaks to another current reality: While Mishra and other IoT experts agree the promise of edge computing and IoT paired together is very real, its actual adoption and implementation (much like 5G) is still in an early phase.
“Fragmentation at the edge – in terms of the type of compute environments, data protocols, connectivity – remains a roadblock,” Mishra says.
Scaling is another critical challenge that grows out of the fragmentation issue.
“Even in the same organization, it can be hard to right-size components given the significant differences that may exist in edge environments regarding hardware platforms, operating systems, connectivity, security, and capacity,” Mishra says. “This remains an interesting space for innovation. If we can overcome this hurdle, it opens massive opportunities to drive value.”
[ Want to learn more about edge and data-intensive applications? Get the details on how to build and manage data-intensive intelligent applications in a hybrid cloud blueprint. ]
Two emerging trends
Mishra shares two other trends worth watching in terms of the edge-IoT pairing that will likely fuel further interest and adoption. The first is video analytics adoption, which Mishra says is a good “quick start” edge-IoT use case.
“Video cameras can be the ultimate sensor and can be added to edge environments as a parallel asset without requiring disruptive instrumentation changes,” Mishra says. “Employing video analytics at the edge can support a variety of use cases because of its flexibility.”
The second is the more tangible emergence of the edge-to-cloud pattern, which is sometimes called the “edge in” approach – versus the “cloud out” approach described above that brings compute and other resources out closer to IoT devices and applications.
“Although the concept of local processing at the edge is powerful, there is a strong need to create a lifecycle effect between edge and cloud for scaling use cases,” Mishra says.
[ How can automation free up more staff time for innovation? Get the free eBook: Managing IT with Automation. ]