Source: ZDNet (original reporting, September 2019)
Edge Computing Is Making a Comeback — and the Linux Foundation Is Leading the Charge
At his keynote address at the Open Networking Summit in Belgium, Arpit Joshipura, The Linux Foundation’s general manager of networking, made a bold prediction: edge computing will overtake cloud computing by 2026. That’s a striking claim given the scale of the cloud industry, but Joshipura’s reasoning is grounded in real technical and market forces that DevOps engineers are already navigating.
The story of computing has always been about where compute power lives. In the mainframe era, it was literally down the hall. The cloud era moved it hundreds of miles away — milliseconds in network terms, but far enough to introduce latency that modern workloads increasingly cannot tolerate. Now, driven by IoT, 5G, autonomous systems, and real-time analytics demands, the pendulum is swinging back toward local compute. That’s edge computing.
What Joshipura Means by “Edge Computing”
Joshipura defines edge computing specifically as compute and storage resources that are five to twenty milliseconds away from the end device or data source. This latency target is the key differentiator from cloud computing and drives most of the architectural choices in edge deployments.
Equally important to Joshipura is the nature of that edge infrastructure. He argues it must be:
- Open and interoperable — not locked to any single vendor’s ecosystem
- Hardware-agnostic — independent of specific silicon or server vendors
- Cloud-agnostic — not tied to AWS, Azure, GCP, or any single cloud provider
- OS-agnostic — capable of running on diverse operating environments
- Universal across use cases — supporting IoT edge, telecom edge, cloud edge, and enterprise edge within a single framework
His stated goal: “Our goal here is to unify all of these.”
LF Edge: The Linux Foundation’s Unifying Edge Initiative
The vehicle for this unification effort is LF Edge, a Linux Foundation umbrella organization designed to consolidate the fragmented edge computing landscape under a common open-source software stack. LF Edge’s purpose is explicitly to prevent the edge ecosystem from splintering into incompatible proprietary silos — the same problem that plagued early cloud infrastructure before projects like OpenStack, Kubernetes, and the CNCF ecosystem emerged.
At the summit, Joshipura announced two new projects joining the LF Edge umbrella: Baetyl and Fledge.
Baetyl (formerly Baidu OpenEdge)
Baetyl is an intelligent edge computing framework originally developed by Baidu and donated to LF Edge. Its core purpose is to extend cloud computing capabilities — data processing, AI inference, and cloud services — down to edge devices, enabling developers to build lightweight, secure, and scalable edge applications without requiring full cloud connectivity at runtime.
Its primary target audience is IoT edge device developers who need cloud-class capabilities closer to the data source. Watson Yin, a Baidu VP, explained the donation decision: “[Baidu] decided to donate Baetyl, the intelligent edge computing framework, to the community, hoping to reciprocate the open-source community while continuously contributing cutting-edge technologies to the global technology ecosystem.”
The motivation follows a pattern familiar in enterprise open source: companies contribute strategically to shared infrastructure projects because a healthy ecosystem accelerates their own commercial products built on top of it.
Fledge (formerly FogLAMP)
Fledge is an open-source framework focused on the industrial edge — factories, plants, and mines where latency, reliability, and safety are critical. Its key use cases include:
- Critical operations monitoring
- Predictive maintenance
- Situational awareness
- Safety systems
Fledge is engineered to bridge Industrial Internet of Things (IIoT) sensors and modern machines with legacy industrial “brown field” systems — equipment that uses proprietary protocols and data formats that predate modern networking standards. It does this through a common set of administration and application APIs shared across both legacy and cloud environments.
Tom Arthur, CEO and co-founder of Fledge’s creator Dianomic Systems, explained the industrial problem clearly: “The LF Edge’s efforts for an open, interoperable framework for the edge is especially needed for the industrial factory, plant, and mine — where almost every brown field system, piece of equipment, or sensor uses its own proprietary protocols and data definitions.”
Why Edge Could Realistically Challenge Cloud Dominance
The cloud is not going away. At the time of Joshipura’s keynote, Gartner had estimated the 2019 public cloud market at approximately $214.3 billion, growing at roughly 17.5% annually. That is an enormous base to overtake. So why does Joshipura believe edge will eventually be larger?
The answer lies in the scale of edge use cases and the environments they span. LF Edge’s vision encompasses industrial, enterprise, and consumer scenarios across complex environments spanning multiple edges and domains — effectively everywhere that compute happens outside a centralized data center. The killer applications driving this shift include:
- Video content delivery — reducing latency for streaming and real-time video
- Autonomous vehicles — requiring sub-millisecond decision loops that cloud round-trips cannot support
- Augmented and virtual reality — demanding low-latency rendering at the edge
- 5G infrastructure — where multi-access edge computing (MEC) is a core architectural component
- Cloud gaming — where input lag directly determines product viability
Two Reasons Edge Computing Has Been Underestimated
Joshipura identified two structural reasons edge computing has not yet received the attention its potential warrants:
1. Fragmentation has worked against it. Edge computing projects have historically operated at cross-purposes, with incompatible standards, competing frameworks, and vendor lock-in preventing the kind of ecosystem momentum that cloud computing achieved. LF Edge exists specifically to solve this problem by creating shared, open infrastructure that the whole industry can build on.
2. The killer apps aren’t fully here yet. Autonomous vehicles, pervasive AR/VR, and 5G-native applications were still emerging in 2019. The mass-market use cases that will drive edge deployment at scale were still maturing. As these technologies reach production readiness, the demand for edge infrastructure should accelerate significantly.
What This Means for DevOps Engineers
For DevOps and platform engineering teams, the rise of edge computing introduces new architectural complexity that pure cloud-native tooling does not yet fully address. Key considerations include:
- Orchestration at the edge: Kubernetes distributions designed for resource-constrained environments (such as K3s and MicroK8s) are gaining traction as the operational model for edge deployments.
- Observability across distributed edges: Monitoring pipelines must extend beyond centralized cloud infrastructure to heterogeneous edge nodes with intermittent connectivity.
- Security posture: Edge devices expand the attack surface significantly; zero-trust network models and hardware-rooted trust become more critical.
- Open standards matter: LF Edge’s push for interoperable frameworks reduces the risk of expensive vendor lock-in at the infrastructure layer.
Whether or not edge computing literally overtakes cloud computing by 2026, the direction of travel is clear: compute is moving closer to the data source, and the open-source infrastructure to support that shift is being built now.