Two industry terms that are widely used, often conflated and poorly defined are ‘edge computing’ and ‘fog computing.’ The purpose of this report is to provide clarity around these two terms, what they signify, key enabling technologies and customer adoption, as well as the market opportunity.
The edge and the fog are two terms used interchangeably in IoT contexts; they lack clear definitions. While both are important and interdependent, they are separate technologies with different considerations and dependencies. This report segments the edge into seven subcategories (species) to better identify the edges being discussed and debated, and distinguishes between the compute and analytics functions performed on edge devices (data plane) and the orchestration and management (control plane) of the fog.
While standard enterprise workloads are gradually transitioning to the cloud, operational technology (OT) and IoT workloads have a number of technical, financial and other legal considerations that dictate workload placement (best execution venue). This has resulted in a continuum of execution venues for IoT data analysis, ranging from being analyzed on the originating device itself, to on-premises compute devices of various levels of scale and sophistication, to off-premises assets in third-party datacenters, hyperscale public clouds, and network operator infrastructure such as multi-access edge computing (MEC). No one edge venue can address all use cases from consumer to military, manufacturing to healthcare. There will continue to be numerous edge venues and points in the topology where the edge data is aggregated for intermediate or regional triage, as well as IoT applications that require aggregation at a central location, such as a public cloud or third-party datacenter.
The end result is a multi-polar and hybrid architecture for IoT deployments that is heavily decentralized and edge-dependent. This decentralized edge requires two types of management and orchestration – one for the hardware and software (VMs, containers, applications) configuration of the devices themselves, and one to manage and apply policy to the application data generated by the connected systems. This has been the original vision of fog computing, which is the large-scale orchestration of devices, enabling software and policy for resources from the edge to the cloud. In contrast, an un-orchestrated IoT deployment, consisting of a hybrid of cloud and edge devices, will be but a small step forward from the inoperable proprietary silos of yesteryear.
This Technology & Business Insight report on edge and fog computing is based on a combination of insights and data gathered through direct interviews with each of the vendors mentioned in the report (with a few exceptions) spanning a wide variety of vertical applications and our analysts' deep experience in the IoT space. Some key findings include:
- Edge computing is the execution venue of choice for IoT analytics in greater than 48% of cases, according to respondents to 451 Research’s Voice of the Enterprise IoT 2019 surveys
- Increasingly robust edge orchestration and management tools are coming to market that have in the past been included under the definition of fog computing, including not only device and application orchestration, but increasingly data policy management as well
- 451 Research estimates the current size of the edge computing market to be nearly $170bn, growing to $651bn by 2024. 451 Research had previously estimated the smaller fog orchestration and management layer to reach $3.7bn in 2019, growing to $18.2bn by 2022.