Running WASM Workloads on Spin with Fermyon Cloud

You push code. It runs in milliseconds. No containers to build. No Kubernetes manifests to debug. No cold starts that kill user experience. That's the promise of WebAssembly on the server, and Fermyon Cloud delivers it with Spin.

Why WASM Changes the Serverless Game

Traditional serverless platforms force you into container overhead. AWS Lambda spins up a microVM. Cloud Run provisions a container. Both add 100-500ms of cold-start latency. WebAssembly modules start in microseconds because they're sandboxed at the runtime level, not the OS level. Spin compiles your Rust, Go, Python, or JavaScript to a .wasm file under 2MB. Fermyon Cloud executes it on a shared Wasmtime runtime with zero container tax.

"
WebAssembly gives you the isolation of containers with the density and startup speed of processes.
— Matt Butcher, Fermyon CEO

Your First Spin App in Three Commands

bash

spin new http-rust my-api
cd my-api
spin build

The spin new command scaffolds a complete project: Cargo.toml, spin.toml, and a handler that returns JSON. spin build compiles to WebAssembly using wasm32-wasi target. No Dockerfile. No docker build wait.

Deploy to Fermyon Cloud

bash

spin login
spin deploy

spin login opens a browser for GitHub OAuth. spin deploy pushes the .wasm module to Fermyon's registry, provisions a route like my-api.fermyon.app, and returns a live HTTPS endpoint in under 10 seconds. Your function scales to zero when idle and handles bursts without configuration.

💡

TipSet <code>spin.toml</code> <code>concurrency = 100</code> to handle 100 simultaneous requests per instance. Fermyon Cloud spins up new instances automatically.

Real-World: Image Resize API

Here's a complete handler that resizes uploaded images using the image crate. Add image = { version = "0.24", features = ["jpeg", "png"] } to Cargo.toml.

rust

use spin_sdk::http::{Request, Response};
use image::ImageFormat;

#[spin_sdk::http_component]
fn handle(req: Request) -> anyhow::Result<Response> {
    let img = image::load_from_memory(req.body())?;
    let resized = img.thumbnail(300, 300);
    let mut buf = Vec::new();
    resized.write_to(&mut std::io::Cursor::new(&mut buf), ImageFormat::Jpeg)?;
    Ok(http::Response::builder()
        .header("content-type", "image/jpeg")
        .body(buf.into())?)
}

Deploy it. POST a 5MB photo. Get a 300px thumbnail back in ~50ms. The binary stays under 3MB. Try that with a container.

When to Choose Spin Over Kubernetes

Factor	Spin on Fermyon Cloud	Kubernetes (EKS/GKE)
Cold start	< 5ms	1-30s
Binary size	1-5 MB	100 MB - 2 GB
Ops overhead	Zero	High
Scaling	Automatic to zero	HPA + cluster autoscaler
Local parity	Exact (spin up)	Approximate (kind/skaffold)

⚠️

WarningWASI preview 1 limits networking to outbound HTTP. No raw TCP, no UDP, no inbound connections except via Spin's HTTP trigger. Check <code>wasi:sockets</code> preview 2 support if you need more.

Next Steps

Install Spin: curl -fsSL https://developer.fermyon.com/downloads/install.sh | bash
Run spin templates list to see Go, Python, JavaScript, and TinyGo options.
Build a cron-triggered job: spin new http-rust --trigger cron.
Explore the Key/Value and SQLite APIs in spin_sdk for stateful serverless.

✦

ℹ️

NoteFermyon Cloud free tier includes 1M requests/month, 100GB-s compute, and custom domains. No credit card required. Deploy your first WASM workload today at cloud.fermyon.com.

Running WASM Workloads on Spin with Fermyon Cloud

Why WASM Changes the Serverless Game

Your First Spin App in Three Commands

Deploy to Fermyon Cloud

Real-World: Image Resize API

When to Choose Spin Over Kubernetes

Next Steps

More Posts

DNA Origami Nanobots Deliver Drugs to Tumors

Rust Async GraphQL Gateway Cuts P99 Latency 40%

NASA CLPS Lunar Lasercomm Delivers 1.2 Gbps Downlink

Build a LoRa Sensor Node That Runs Six Months on AA Cells

Nvidia Thor SoC Cuts Robotaxi Compute Cost 40 Percent

Integrate CRISPR Guide RNA Design Into Your Bio Pipeline

Use DuckDB to Cut Your Analytics Pipeline Costs in Half

eBPF Runtime Security Cuts Container Breach Detection to Seconds

On-Device ML Autofocus Cuts Camera App Latency to Under 50ms

eBPF XDP Load Balancing Cuts Kubernetes Latency 40 Percent