Edge computing moves compute and storage closer to where data is generated — factories, retail locations, hospitals, cell towers — rather than centralizing everything in distant cloud regions. Latency, bandwidth, and data sovereignty requirements drive this shift.
Why Edge Matters
When milliseconds matter (industrial control, real-time video analytics, autonomous vehicles) or when bandwidth to the cloud is constrained (remote locations, offshore platforms), processing at the edge avoids the round-trip to a cloud data center.
Cloud Provider Edge Offerings
AWS Outposts
AWS Outposts delivers AWS infrastructure — servers, storage, and networking — to your on-premises location. It runs native AWS services (EC2, EKS, RDS, S3 on Outposts) using the same AWS APIs and console.
- Outposts Rack — full 42U rack of AWS hardware
- Outposts Server — 1U or 2U for space-constrained locations
- Managed and monitored by AWS; connectivity back to parent AWS region required
Azure Stack
Microsoft offers multiple edge products:
- Azure Stack Hub — full Azure services in a disconnected or partially connected environment
- Azure Stack Edge — hardware appliance with ML inference capabilities at the edge
- Azure Arc — manage on-premises and multi-cloud resources from Azure control plane without Azure hardware
Google Distributed Cloud
- Hosted — Google-managed hardware at co-location facilities
- Edge — customer-managed hardware in air-gapped environments
- Runs GKE and other Google services
OpenNebula at the Edge
OpenNebula is well-positioned for edge deployments. Its lightweight architecture — unlike OpenStack — runs efficiently on small clusters. The OpenNebula Edge Cloud model provisions cloud resources at remote locations and federates them into a unified management plane, a direct evolution of the distributed computing research done at UCM.
Edge vs. Fog vs. CDN
- CDN edge — static content caching close to users (CloudFront, Cloudflare)
- Fog computing — intermediate nodes between devices and cloud
- Edge computing — compute at or near the data source
Use Cases
- Manufacturing — real-time quality control, predictive maintenance
- Retail — local inventory management, in-store analytics
- Healthcare — medical imaging analysis without sending patient data off-site
- Telecommunications — 5G network function virtualization at base stations