CONCEPT Cited by 1 source
Edge computing¶
Definition¶
A distributed computing paradigm where computation and data storage are brought closer to the sources of data — physically located at or near the point of use rather than in centralized cloud data centers. In system design, edge computing addresses three primary concerns: latency (network round-trips eliminated), connectivity (operation despite intermittent or absent network), and data locality (sensitive data stays on-premises).
Key trade-offs¶
- Capability vs. proximity: Edge hardware is constrained relative to cloud; model/workload selection is hardware-driven.
- Freshness vs. independence: Updates require deployment cycles; edge runs on stale artifacts between syncs.
- Fleet management complexity: Each edge device is an independent failure domain with its own observability, update state, and hardware profile.
- Security perimeter shift: At the edge you own the full stack — physical access, encryption, network segmentation.
Seen in¶
- sources/2026-07-22-aws-architecting-offline-first-generative-ai-applications-for-edge-deployments — manufacturing/industrial AI inference on edge devices (NVIDIA T4/Jetson Xavier GPUs) using concepts/offline-first-architecture with IoT Greengrass as the cloud-to-edge deployment bridge.
Related¶
- concepts/offline-first-architecture — the design axiom for edge AI systems
- concepts/edge-inference — the specific ML inference case
- systems/aws-iot-greengrass — AWS's edge deployment orchestrator