No Code Request Routing

Modern applications don't live at a single origin. You've got API servers, static assets, auth services, media pipelines, each needing traffic to route correctly. Getting there meant complex configuration: multiple domains, custom VCL, compute code sitting in front of your origin, or the use of service chaining.

Request Routing can simplify this for you.

One Domain, Multiple Services

With Request Routing, you define path-based rules on any domain in our Domain Management system, and we send each request to the right service automatically, no custom code required.

Host your marketing pages on a static CDN, your API on a containerized backend, and your media files on object storage, all under one domain. Route /api/* to your app server, /assets/* to S3, /auth/* to your identity provider, clean, fast, and entirely declarative.

Use Cases

• Microservices Orchestration: Route different API versions to specific legacy or modern services.

• Geographic Localization: Direct users from specific countries to localized origin servers to reduce latency.

• A/B Testing & Canary Deploys: Use header-based matching to route a percentage of traffic to staging or beta services.

How It Works

Route configurations live under the Domains top-level navigation in your dashboard. You can create sophisticated rules using three primary condition types. Our first release will only support headers, but the other condition types are coming soon. 

• Header: Match based on HTTP headers like API versions or User-Agents.

• Geo: Route traffic based on the user's geographic location, such as country codes.

• IP: Filter and route based on specific IP addresses or ranges.

For this release, you can check whether a header value starts_with, ends_with, equals, or contains a given string. Every path can define multiple rules, and each rule has an associated action. The action determines what happens when/if a condition is matched. For the initial release we're providing a 'service' action that takes a Service ID as its value (we have plans for other action types, coming soon).Every path requires a default action, a fallback route for when no rules match.

Here's a simple example of routing traffic based on the User-Agent request header:

Route configuration:
  Path: /example
    Rule:
      Condition: header User-Agent starts_with "Mozilla/5.0"
      Action: route to service-A
    Rule:
      Condition: header User-Agent starts_with "Mozilla/4.0"
      Action: route to service-B
    Default: route to service-C

The system evaluates requests to /example top-to-bottom. The first matching rule wins; anything that does not match any condition falls through to the default.

The Draft-to-Deploy Workflow

Request Routing ensures safety and increases developer velocity by following a streamlined "Draft-to-Deploy" workflow:

1. Draft: When you start an edit, we automatically create a draft by cloning your active configuration.

2. Configure: Add paths and define rules. Each route configuration supports up to 10,000 paths, 20 rules per path, and 5 conditions per rule.

3. Preview: Use the Structured Diff API endpoint or UI to review exactly what changes you made before going live.

4. Activate: Deploy your changes to the edge with a single call. Strict validation ensures every path has a fallback rule.

5. Rollback: We store the last five active versions, allowing for near-instant rollbacks to a known-good state.

Get Started

Request Routing is available now on all plans. Head to your dashboard to create your first configuration, or read the documentation for rule syntax, wildcard matching, and advanced options.

Building something interesting with it? Share it on the Fastly community or drop us a line at Fastly Support.