Monitoring edge and fog computing devices

Blog: Monitoring edge and fog computing devices

Edge computing and fog computing are technological advancements gaining traction in a hyper-connected world. Being close to the source, edge computing enables data collection and processing at the fastest possible speeds.
Instead of sending all the data to a remote cloud location through the internet with latency, edge devices store and process most of it onsite and pass the heavy lifting to the central cloud to achieve the quickest turnaround. In a way, edge computing is like the spinal cord handling emergency responses without sending them to the brain. Research firm IDC forecasts spending on edge solutions will rise rapidly through 2025 to reach about $274 billion.
Edge computing helps achieve unprecedented response rates, contextual data handling, bandwidth savings, the lowest latencies, and the highest possible uptime. By integrating intelligence and storage within devices, edge computing creates a distributed, decentralized, and capable world that does not cease when a network fails.

What is fog computing?

An extension of edge computing, fog computing is an intermediary stage between the edge and the cloud, offering the best of both worlds. Like how fog hovers close to the ground, fog computing is an apt name for a decentralized IT infrastructure closer to devices, delivering excellent last-mile connectivity and autonomy. 

Fog computing boosts the capabilities of edge devices by bringing the power of the cloud as physically close as possible to devices; it handles additional data processing, storage, and apps while compensating for capacity limitations and technical restrictions of the cloud by being closer to the edge.

Monitoring needs

Post-lockdown businesses need low-latency high-speed business networks with integrated compute layers for robust, collaborative applications. In a world of hybrid work and collaboration, edge and fog services are gaining ground as essential ingredients for the success of the Industrial Revolution 4.0, which is the popular name for hyper-connected and supremely flexible and capable manufacturing.
The numbers are showing. More than half a million industrial robots were installed in 2021 worldwide, up 31% from 2020 and double from six years ago. Subsequently, as the edge and fog computing market grows, newer challenges emerge, and chief among them is the need to monitor them well.

Despite the promise and push, significant barriers prevent the mass adoption of edge and fog computing models. These barriers include a lack of mature software systems, evolved edge and fog computing architecture, and the development of capable monitoring solutions to deal with all emerging edge and fog complexities at scale.

Unique challenges

There are three types of edges to the endpoints they serve: the client edge, the network edge, and the data center or cloud service edge. 

Client edges are the edges between IT domains or network interfaces, a network edge is an edge between different IT domains, and a data center or cloud service edge is the app server that first receives the traffic from clients. Edge monitoring is essential across all these edges to ensure the best visibility and troubleshooting capabilities during downtime.
Compared to centralized IT infrastructure that is easier to monitor, edge and fog computing infrastructure is tougher to observe, monitor, and manage due to its varied and distributed endpoints and potential blind spots in observability. A single edge node that goes unmonitored could stop access for a remote worker, a shopper, or even a critical system whose functioning depends on a specific edge's data relay.

It is also possible that hackers could use edge computing and fog computing devices as entry points to attack central systems because of the physically diversified distribution of edge and fog computing devices. Experts say that a sound monitoring strategy with prompt remediation and information relay routines would be necessary to ensure the best security for edge and fog devices.

Overcoming limitations

Monitoring edge nodes is a complex challenge. Traditional agent-based monitoring is not feasible because edge nodes do not allow external software installations. In addition, some edge nodes are too small to support traditional monitoring agents, such as those that use processors like Raspberry Pi or Arduino. These nodes do not favor TCP/IP protocols, which are the foundation of traditional monitoring agents.

Some edge computing providers manage and monitor the edge nodes to address these challenges. These providers supply telemetry data for monitoring solutions to consume directly. For example, AWS CloudFront provides a telemetry data feed that can be used to monitor edge nodes. Content delivery network services such as Cloudflare also supply logs that can be used to mine telemetry data.

Emerging ideas for monitoring edge nodes include orchestrating monitoring agents on fog computing clusters. Fog computing clusters are distributed computing infrastructures closer to the network's edge than traditional cloud computing infrastructure. Orchestrating monitoring agents on fog computing clusters allows for more efficient and effective monitoring of edge nodes.

In addition to orchestration, future monitoring solutions are expected to evolve to handle the heterogeneity of network endpoints and fast-changing schemes for devices. These solutions must also be capable of cutting noise and false alerts to gain quality monitoring data. All such emerging monitoring solutions must rethink traditional cloud monitoring methods to suit the unique aspects of fog and edge computing realms.

Vast potential

Fog computing orchestrations need monitoring to gain actionable data to feed decision-makers through appropriate channels. Since fog environments may not have dedicated data processing facilities, IT teams should think beyond agent-based monitoring systems and discover newer methods to take cognizance of the readiness, functioning, and security of fog systems and edge devices. 

The top monitoring challenges are: How will issue alerts be sent? What are the best metrics to track? How much computing power is necessary for monitoring each edge and fog device? How would all the real-time information be relayed? Would it be pushed or pulled?

Edge and fog computing holds vast potential to transform a wide range of industries, including medical systems, smart cities, manufacturing parks, and military applications. Today, industry collectives like the Open Fog Consortium are working to develop the reference architecture for edge and fog computing deployments.

As edge and fog computing continue to evolve, it is essential to keep asking relevant questions to ideate and develop monitoring solutions tailored to these technologies' unique requirements. Only by transcending today's monitoring limitations can we ensure that edge and fog computing deployments are reliable, efficient, and secure.

Comments (0)