Edge Computing
Definition
Processing data closer to its source rather than relying solely on centralized data centers, reducing latency and improving response times for real-time
Use Cases
- Netflix: Low-latency video streaming and reduced backbone traffic by serving content close to viewers — Deployed the Netflix Open Connect CDN with caching appliances installed at or near ISP networks, keeping popular content at the network edge rather than pulling it from centralized data centers (Improved streaming performance and reliability for end users while reducing transit costs and congestion for both Netflix and participating ISPs)
- Cloudflare: Running security, performance, and application logic close to users (e.g., DDoS protection, caching, and edge application execution) — Operates a global edge network and provides serverless execution at the edge (Cloudflare Workers) so code can run in data centers near end users instead of a single central region (Lower latency for end users and faster response times for applications, with improved resilience by distributing traffic handling across many locations)
- John Deere: Precision agriculture: processing machine and sensor data in the field to support near-real-time decisions — Uses on-machine and near-machine computing to process sensor/camera/telemetry data locally where connectivity can be limited, sending selected data to the cloud for longer-term analytics and fleet management (Faster in-field insights and reduced dependence on continuous connectivity, enabling more responsive operations and better use of collected data)
Provider Equivalents
- AWS: AWS Outposts / AWS Wavelength / AWS Local Zones / AWS IoT Greengrass
- Azure: Azure Stack Edge / Azure Arc / Azure IoT Edge
- GCP: Google Distributed Cloud (Edge) / Anthos
- OCI: OCI Roving Edge Infrastructure
Frequently Asked Questions
- What's the difference between Edge Computing and Cloud Computing?
- Cloud computing runs workloads in centralized provider data centers (regions). Edge computing runs some processing closer to where data is created—on devices, gateways, factories, retail stores, or telecom locations. Edge is used when you need very low latency, local autonomy during network outages, data residency constraints, or to reduce bandwidth by filtering/aggregating data before sending it to the cloud.
- When should I use Edge Computing?
- Use edge computing when (1) latency must be extremely low (robotics, AR/VR, industrial control, autonomous systems), (2) connectivity is unreliable or expensive (remote sites, ships, mines), (3) you generate lots of data and only need to send summaries to the cloud (video analytics, IoT telemetry), (4) privacy or regulatory rules require local processing (healthcare, retail video), or (5) you need local resilience so operations continue even if the cloud link fails.
- How much does Edge Computing cost?
- Costs vary widely based on where the edge runs and how it’s managed. Common cost drivers include: edge hardware (servers, gateways, ruggedized devices), software licensing or managed services (Kubernetes/edge runtimes, device management), connectivity (cellular/5G, private links, bandwidth), operations (remote monitoring, patching, on-site support), and data transfer to the cloud. Some offerings are priced like cloud resources (per hour/per vCPU/per GB), while others require purchasing or leasing hardware plus support contracts.
Category: emerging
Difficulty: advanced
Related Terms
See Also