Client side compute and edge compute are converging. Web apps are transitioning where data is processed closer to the user—a move inclusive of not o

Whither Serverless Compute? or Why the Cloudflare-PartyKit Acquisition Matters

submited by

Style Pass

2024-04-26 18:30:34

Client side compute and edge compute are converging. Web apps are transitioning where data is processed closer to the user—a move inclusive of not only compute on the device using client side rendering engines like V8, Google’s JavaScript and WebAssembly engine, but also the so-called “edge.” As Stephen O’Grady has so eloquently explained, despite “dozens or perhaps hundreds of technology vendors at present hav[ing] substantial open investments in marketing edge branded products,” the term “edge” is rife with ambiguity. However, beyond exasperating critics if “edge” accomplishes nothing else, it does the important work of signifying servers that are geographically closer to the user than cloud data centers historically have been.

This trend, which AI workloads will only serve to amplify, is moving serverless compute closer to the user for performance reasons—a transition which has repercussions for packaging, runtimes, databases, and abstractions in general. Serverless compute (an idea that is itself fraught with ambiguity, as Rachel Stephens notes) is where all this convergence around the location of data processing, which has migrated away from local data centers (colocation, rackspace), and toward the cloud and the devices themselves (browsers, IoT), whether you want to use the term “edge” or not.