Edge Computing

The traditional Cloud model is simple: In most cases, this is fine. With this approach, we assume that latency is not important, bandwidth is cheap, and volume is small. Clearly, these assumptions do not always hold. What if, for example, we have a cloud system, that requires millisecond latency and has a high load of data. Edge Computing is a System Design principle that solves this problem by simply moving computation closer to the servers/data.

The edge is any cluster of nodes closer to the compute servers than the current user. This can be a CDN, a data center, or even just a raspberry pi next to a sensor. Edge Computing makes it clear that centralization in the Cloud is a real cost and may be too expensive for some applications. Typically, edge workloads are fast, deterministic, resource constrained, and statelessOr just barely stateful. It is more "decision logic" than main responsibility. These workloads are typically executed per request, sandboxed, and auto-scaledMaking it a great candidate for Web Assembly