Fastly, the cloud computing company providing solutions for edge cloud networks and content delivery networks, recently announced new features for their serverless compute environment Compute@Edge. The new release introduces new CLI functionalities, Terraform API support and additional language support.
Introduced as a beta a year ago, Compute@Edge is now generally available and offers support for Rust and AssemblyScript, a variant of TypeScript that produces WebAssembly binaries. In an article discussing the most challenging aspects of serverless and how Fastly plans to address them, MJ Jones, principal product manager at Fastly, explains the main benefits of Compute@Edge:
Serverless promises a huge advantage: you no longer need to create and dedicate a specific amount of server resources for your applications (...) but not all serverless solutions are created equal. (...) Cold starts, regional latency, and observability are among the most commonly perceived challenges. But Compute@Edge, our serverless computing environment built on WebAssembly and Lucet, ushers in a new generation of serverless.
According to the author, the service reduces significantly regional latency running logic simultaneously across servers around the world and providing code execution startup times measured in microseconds, significantly lower than for other serverless platforms. Jeremy Daly, chief technology officer at AlertMe and AWS Serverless Hero, is impressed by the new platform and comments in the Off-by-none weekly serverless newsletter:
Remember all those times I told you to pay attention to Fastly? Well, they not only announced major enhancements to their Compute@Edge solution but also acquired part of the WebAssembly (Wasm) team from Mozilla. If you’re not thinking about Wasm-on-the-server, you should be. It’s likely the future of compute at the edge.
As reported in the Bytecode Alliance: one year update, Fastly has recently acquired part of the WebAssembly team from Mozilla to expand their work on Lucet, the open-source WebAssembly runtime and compiler that powers Compute@Edge.
Among the use cases for the edge platform, Fastly suggests waiting room tokens to limit the rate at which users can start new sessions, machine learning inference processing, IoT data stream summarization and serverless applications that require processing capabilities at the entry points, reducing the logic in centralized data centers. The Developer Hub provides recipes, templates and server kits for developers who want to try edge computing using Computer@Edge.
Waiting room implemented at edge. Source: https://developer.fastly.com/solutions/patterns/waiting-room/
Fastly is not the only company expanding edge computing deployments and options: Cloudflare Workers is Cloudflare's service to run serverless code across the globe. According to Jason Warner, chief technology officer at GitHub, the two companies have a significant lead on the largest cloud providers:
I believe the most interesting cloud tech is at the edge and the large cloud providers are missing it. Both Fastly and Cloudflare have room to run here and it’s a wide open field full of ponies and rainbows.
AWS offers Lambda@Edge as a feature of Amazon CloudFront to build more responsive applications and analytic services closer to end-points. Microsoft introduced earlier this year Azure Edge Zones, while Google Cloud’s most recent announcements include Anthos at the Edge.
Peter Bourgon, principal engineer at Fastly, presented earlier this year at QCon the talk Infinite Parallel Universes: State at the Edge that explains how edge cloud networks and content delivery networks work.