Edge Computing: What Is It?
The foundational principle of edge computing is that latency sensitive applications need to be as close to the consumer as possible. Proximity to the user is critical to ensure minimal response times. While the concept of edge computing is simple to describe, it is very difficult to execute. This is especially true of the emerging category of hybrid edge computing.
At its most basic, edge computing can be reasonably viewed as a repackaging of traditional on-premises computing. The concept of processing data as close as possible to where it was collected wasn’t considered novel or controversial until cloud computing became a thing; workloads started moving out across the Internet, and started being restricted to a handful of large centralized data centers.
Unfortunately for the enterprise, public clouds aren’t the solution to all things IT. One primary problem with public clouds is latency. Try as we might, we haven’t yet managed to communicate faster than the speed of light, and the distance between physical premises where data is generated and the data centers of the large public cloud providers can simply be too far apart.
Each workload has its own latency sensitivity. Latency is the combination of the time it takes data to get from the data generation source to the data processing application, the time it takes the relevant application to process the data, and the time it takes the result to travel to the destination.
As a general rule, latency higher than 100msec, or 1/10th of a second, is concerning. Latencies of 300msec or higher usually mean that an application cannot be considered to serve data in real time. As applications get more complex and require more and more data sources, this problem multiples. Applications within the same data center enjoy a sub-ms or 1-2ms range, where metro traffic between facilities may be slightly higher, but well below 3-5ms.