Fastly Compute@Edge

Fastly Compute@Edge
Name	Compute@Edge
Developer	Fastly
Released	2020s
Operating system	V8 isolates
Language	Rust, JavaScript, TypeScript, WebAssembly
License	Proprietary

Fastly Compute@Edge

Compute@Edge is a distributed serverless edge computing platform provided by Fastly that runs application logic at the network edge. It enables developers to deploy WebAssembly-based services close to end users leveraging a global content delivery network operated by Fastly, integrating with ecosystems from Mozilla to Cloudflare Workers-adjacent technologies and competing with offerings from Amazon Web Services and Google Cloud Platform. The platform is positioned for low-latency HTTP processing, API orchestration, and content personalization across Fastly's points of presence alongside partners such as Akamai Technologies and Microsoft Azure.

Overview

Compute@Edge is designed to move compute from centralized Amazon Web Services regions and Google Cloud Platform zones to Fastly's distributed edge, enabling HTTP request and response manipulation, caching logic, and real-time transformation. The platform is informed by prior work in edge runtimes like Cloudflare Workers and influenced by standards from World Wide Web Consortium and the WebAssembly community. Fastly markets Compute@Edge for use by organizations including digital publishers, streaming platforms, and e-commerce sites that also work with vendors such as Shopify, The New York Times, and GitHub.

Architecture and Components

The architecture centers on a global network of points of presence similar to architectures used by Akamai Technologies and Cloudflare, with a control plane for service deployment and a data plane executing workloads in lightweight isolates. Key components include a developer CLI inspired by tooling from Mozilla and Google Chrome's developer tools, a build pipeline producing WebAssembly artifacts compiled from languages like Rust (programming language) and TypeScript, and logging/observability integrations compatible with Datadog and New Relic. Origin shield and caching behavior interoperate with Fastly's Varnish-based heritage and edge dictionaries influenced by designs from Nginx-based reverse proxies.

Programming Model and Languages

Compute@Edge supports writing functions in languages that compile to WebAssembly, notably Rust (programming language), TypeScript, and JavaScript via toolchains similar to AssemblyScript and LLVM-based flows. The programming model emphasizes event-driven handlers for HTTP requests, mirroring patterns in Service Workers and runtimes such as Cloudflare Workers, while exposing Fastly-specific APIs for caching, streaming, and request modification. Developers often combine SDKs from ecosystems like npm and build systems from Webpack or esbuild alongside package managers such as Cargo (package manager) for Rust.

Security and Sandboxing

Security relies on isolation through WebAssembly-based sandboxing akin to approaches advocated by Mozilla and the Linux Foundation's initiatives, with runtime constraints similar to security models in Google Chrome's V8 isolates and WebKit. The platform enforces capability-based APIs to minimize lateral movement and integrates with identity providers like Okta and Auth0 for access control, while observability is compatible with audit tooling from Splunk and Sentry (software). Isolation benefits echo research from Intel and ARM Holdings on trusted execution environments, though Compute@Edge operates at the software-isolation layer.

Performance and Scaling

Compute@Edge targets millisecond-scale latency improvements by executing at Fastly's edge locations, comparable in objective to performance goals of Akamai Technologies and Cloudflare. Scaling is automatic across Fastly's points of presence, with cold-start characteristics influenced by WebAssembly startup times studied in work from Mozilla Research and benchmarking frameworks used by SPEC. Integration with HTTP/2 and HTTP/3 features like QUIC, developed by contributors to IETF, further reduces round-trip times for global audiences including users served via networks from AT&T or Verizon Communications.

Use Cases and Integrations

Common use cases include A/B testing and feature flags for publishers such as The Washington Post, dynamic image and video optimization for streaming services like Netflix-adjacent workflows, API gateway logic for microservices architectures used by enterprises such as Stripe or Square (financial services), and security middleware for web applications alongside services from Cloudflare and Akamai Technologies. Integrations span CI/CD platforms such as GitHub Actions and GitLab and observability stacks from Datadog, New Relic, and Splunk.

History and Development

Compute@Edge evolved from Fastly's heritage in Varnish-based CDN technology and responses to emergent edge computing paradigms promoted by Cloudflare Workers and research from Mozilla Research and the WebAssembly community. Fastly announced and iterated the service during the 2020s amid industry shifts involving major cloud providers including Amazon Web Services and Microsoft Azure, with development informed by open-source projects such as Varnish (software) and standards efforts led by W3C and the IETF. Ongoing development reflects competitive dynamics with companies like Cloudflare and collaborations with observability and security vendors including Datadog and Splunk.

Category:Edge computing