Chapter 36: Edges

The edge of the network is where the planetary scale computer meets its users. Edge computing moves computation and data closer to users — into local data centers, points of presence (PoPs), cell towers, and even end-user devices. This reduces latency, decreases backbone traffic, and enables applications that require real-time responsiveness.

Content Delivery Networks (CDNs) are the most common form of edge infrastructure. A CDN caches static content (images, videos, scripts) at hundreds of edge locations around the world, serving users from the nearest location. This transforms a 200-millisecond cross-continent fetch into a 5-millisecond local cache hit.

Beyond caching, edge computing can run application logic close to users. Edge functions (similar to cloud functions but running at edge locations) can handle authentication, personalization, and request routing without a round trip to the origin data center. This blurs the line between the edge and the data center, creating a continuum of compute locations from the user's device to the central facility.

The challenge of edge computing is managing consistency across many small locations. Each edge location is a potential source of stale data or conflicting state. The techniques we have studied — caching with TTLs, eventual consistency, and consensus for critical data — all apply at the edge, but the trade-offs shift toward availability and low latency.

Our three-region deployment (SFO, NYC, AMS) is a practical form of edge-like distribution — not hundreds of locations, but enough to cover major population centers with low-latency access. Chapter 24: Geo Replication shows how the full stack runs in each region with asynchronous cross-region replication.