1. Home
2. Guides

The Fastly Compute platform is an advanced edge computing system that runs your code, in your favorite language, on our global edge network. Security and portability are provided by compiling your code to WebAssembly. We run your code using Wasmtime.

With the Compute platform you have access to Fastly edge primitives like data stores, dynamic configuration, and real-time messaging. The platform is completely language agnostic at every layer of the stack, and you can include dependencies from your preferred package registry.

Whether it's authentication, personalization, geofencing, SEO, observability, templating, APIs, or even your entire application, whatever you can conceive, you can probably run on the Compute platform.

Getting started

To use the Compute platform, you must create a Fastly account if you don't already have one.

Generate an API token

To authenticate the Fastly CLI, create an API token for your account. Make sure the token has a global scope, and make a note of the token.

Set up the development environment

Install the Fastly CLI, and then verify that everything works by running fastly version. For example:

$ fastly version
Fastly CLI version vX.Y.Z (abc0001)
Built with go version go1.20.1 linux/amd64

Configure the CLI to act on your behalf using the API token you created:

$ fastly profile create

This will store your API token credential in a configuration file and remember it for subsequent commands. There are other ways of authenticating the CLI if you prefer.

Choose a language to use

We provide SDKs for multiple languages that can compile to Compute-compatible WebAssembly packages. Here is a comparison of SDK-exposed features that can differ among languages:

Current SDK version	0.11.2	3.28.0	1.3.3
Equally supported	Geolocation, Auto decompression, Dynamic compression, Environment variables, Cache override, Real time logging, KV stores, Config stores, Secret stores, Fanout, Dynamic backends, Simple cache, Readthough cache
Core feature differences
WebSockets passthrough ℹ️	✅	-	-
Mutual TLS for origin fetch	✅	-	-
Core cache interface ℹ️	✅	✅	✅
Readthrough cache - custom cache keys ℹ️	✅	✅	-
Readthrough cache - customize cache behavior ℹ️	✅	-	-
Pre-release feature differences
Purging see note below	✅	-	-
Backend health	✅	✅	-
Device Detection	✅	✅	✅
Edge rate limiting ℹ️	✅	✅	✅
	More about Rust »	More about JavaScript »	More about Go »

We recommend the use of official Fastly SDKs (available for Rust, JavaScript, and Go), but you can also create your own for a language of your choice, or use a community-created SDK. Learn more about custom SDKs.

Install language tooling

To build your project, the Fastly CLI requires your local toolchain to be available and up to date. To install the toolchain for your chosen language, follow these instructions:

$ curl --proto '=https' --tlsv1.2 -sSf https://sh.rustup.rs | sh -s -- -y --default-toolchain 1.82.0

$ rustup target add wasm32-wasi --toolchain 1.82.0

$ rustup toolchain add 1.82.0

$ rustup target add wasm32-wasi --toolchain 1.82.0

$ rustup default 1.82.0

Create a new project

Type fastly compute init to scaffold a new Fastly Compute project. The CLI will generate source code for your project in the current working directory.

You will be asked to choose a language and then a starter kit to use as the starting point for your project. There are many starter kits available. If this is your first project, we recommend choosing the default starter for your chosen language.

$ mkdir example && cd example
$ fastly compute init

Follow the prompts to configure your new project:

Creating a new Fastly Compute project.

Press ^C at any time to quit.

✓ Validating directory permissions

Name: [example]
Description: My new Rust-based app!
Author (email): jo.test@example.com
Language:
[1] Rust
[2] JavaScript
[3] Go
[5] Other ('bring your own' Wasm binary)
Choose option: [1] 1
Starter kit:
[1] Default starter for Rust
    A basic starter kit that demonstrates routing, simple synthetic responses and
    overriding caching rules.
    https://github.com/fastly/compute-starter-kit-rust-default
[2] Empty starter for Rust
    An empty starter kit project template.
    https://github.com/fastly/compute-starter-kit-rust-empty
...

Choose option or paste git URL: [1]

At the prompts, provide details about the service you're creating:

• Name: Enter a name for your service. By default, the CLI uses the name of the current directory.
• Description: Optionally enter a short description for your service, or leave it blank.
• Author: Enter an email address or accept the default, which is the email associated with your API token.
• Language: Select the language you want to use to write your code.
• Starter kit: Choose from the available starter kits or press enter to accept the default. You can also enter the URL of a GitHub repository to use a compatible template that's not in the list of starter kits.

A local development environment and a fastly.toml package manifest will be generated based on your selections:

✓ Fetching package template
✓ Reading package manifest
✓ Setting package name in manifest to "example"
✓ Setting description in manifest
✓ Setting authors in manifest to 'jo.example@example.com'
✓ Setting language in manifest to 'rust'
✓ Saving manifest changes
✓ Initializing package

Initialized package example to:
    /home/jo/repos/example

To publish the package (build and deploy), run:
    fastly compute publish

SUCCESS: Initialized package example

Compile to WebAssembly

You already have a complete, functional Compute project at this point. A default starter kit will run fine without any code changes, but if you want to dive in to the code right now, check out the language-specific guides on Rust, JavaScript or Go.

To run the Compute project on Fastly, it needs to be compiled into a WebAssembly module that can be packaged for Fastly use. The CLI runs the appropriate compiler for the language you're using by executing the scripts.build command defined in your project's fastly.toml file, and generates the necessary WebAssembly binary.

$ npm install

Type fastly compute build when you're ready to build:

$ fastly compute build
✓ Verifying package manifest
✓ Identifying package name
✓ Identifying toolchain
✓ Running [scripts.build]
✓ Creating package archive

SUCCESS: Built package (pkg/<name>.tar.gz)

Deploy to a Fastly service

To deploy the project to Fastly, run fastly compute deploy.

If you already have a Fastly service you want to deploy the package to you can add the service ID into the service_id property in fastly.toml before running the deploy command; otherwise the CLI will lead you through creation of a service automatically.

$ fastly compute deploy

There is no Fastly service associated with this package. To connect to an existing service
add the Service ID to the fastly.toml file, otherwise follow the prompts to create a
service now.

Create new service: [y/N] y

Service name: [example]

✓ Creating service

Domain: [random-funky-words.edgecompute.app]

Backend (hostname or IP address, or leave blank to stop adding backends):

✓ Creating domain 'random-funky-words.edgecompute.app'
✓ Uploading package
✓ Activating service (version 1)
✓ Checking service availability (status: 200)

Manage this service at:
    https://manage.fastly.com/configure/services/PS1Z4isxPaoZGVKVdv0eY

View this service at:
    https://random-funky-words.edgecompute.app


SUCCESS: Deployed package (service PS1Z4isxPaoZGVKVdv0eY, version 1)

At the prompts, provide details about the service you're creating:

• Domains: Press enter to accept the automatically generated domain name. We'll take care of the TLS and DNS and give you a working domain immediately. It's a good idea to accept the default now and add more domains to the service later.
• Backends: If you want to configure an origin server for your service, enter a valid hostname, or an IPv4 or IPv6 address. Alternatively, leave the prompt blank to create a service with no origin server. Such services cannot forward requests to your servers, but they can respond to client requests with a response composed within your Compute package.

The output includes a link to manage your service in the Fastly web interface. You can use this link to perform more complex configuration of the service, such as adding or removing domains, changing origin server settings, and setting up logging endpoints.

After the command completes, it may take up to a minute before the new service begins handling requests, and if you're replacing a previous version of the package, the earlier version may continue to handle requests for that period.

Test and celebrate

Open a browser and head over to the domain that was assigned to the service, such as random-funky-words.edgecompute.app, and you should see a working site. If you used the default starter kit, it will look something like this:

Add your own domain

Now that your working project is deployed, you may want to assign a domain name to it. We have a lot of TLS options. As a starting point, it's a good idea to create a TLS Subscription so that Fastly will generate a certificate for you.

• Setting up a custom domain with a TLS subscription.

Go further

If you would like to learn more about how the Compute platform integrates with your preferred language, read our getting started guides for each of our officially supported languages:

• Using Rust
• Using JavaScript
• Using Go

If you are already familiar with VCL and want to migrate an existing VCL service to the Compute platform, try the VCL to Compute migration guide, which shows examples of the most common use cases in VCL and how to implement the same functionality using the Compute platform's supported languages.

Or maybe you have an idea in mind and want to experiment with examples? Try cloning a starter kit, copy and pasting code from our code examples, or following a tutorial.

Limitations and constraints

Compute services accept client connections on port 443 only. This is different from VCL services, which support client connections on port 80.

The following default limits apply to Compute services. If you need to exceed any other limits, please speak to your Fastly contact or contact Fastly support:

An 'execution' refers to a single instance of a Compute program being executed, normally in response to a client HTTP request. Separate limits also apply to the use of log tailing with Compute services.