Edge computing is the practice of doing computation close to where data is produced and consumed, rather than sending all of it across the network to a centralized cloud and waiting for an answer. The “edge” is the part of the network nearest the physical world: the sensors, the devices, and the local gateways that sit between them and the wider internet. RFC 9556, the IETF’s 2024 document on IoT edge challenges and functions, defines edge computing as distributing “computing, storage, control and networking functions closer to the users along a cloud-to-thing continuum,” so that data is “stored, processed, analyzed, and acted upon close to the data sources.”
The motivation is partly about latency. Cloud round trips take tens or hundreds of milliseconds, which is fatal for control loops that must react in real time. RFC 9556 notes that “industrial robots and motion control systems have use cases for cycle times in the order of microseconds,” timescales for which a trip to a remote data center is simply not an option. Processing locally lets a machine decide and act before a cloud could even receive the request.
Bandwidth is the second driver. A camera or vibration sensor can generate far more raw data than is practical to upload continuously, and many deployments have no fast, reliable, or affordable uplink. RFC 9556 states plainly that “minimizing reliance on high-bandwidth connectivity is a requirement,” because maintaining continuous connectivity is “unjustifiable and impractical for most IoT applications.” Filtering, aggregating, or summarizing at the edge means only the useful results travel onward.
Privacy and resilience round out the case. By “computing sensitive data closer to data generators,” edge architectures can keep raw, identifiable data out of the cloud entirely, and they reduce what a “passive observer” can learn by watching device-to-cloud traffic. A local edge node can also keep working through an internet outage, since the logic it needs is already running on site.
RFC 9556 places edge computing on a continuum rather than treating it as a single tier, and it describes “fog computing” as “multi-tiered edge computing,” meaning “several layers of compute infrastructure between end devices and cloud services.” In practice an architecture may run inference on the device, aggregation on a nearby gateway, and heavier analytics in a regional facility, with the cloud reserved for long-term storage and global coordination. Edge computing did not replace the cloud so much as complement it, and it is now a foundational pattern for IoT, real-time control, and latency-sensitive applications.