Cloudflare Workers serving 15+ sites from one deployment.

The routing and middleware layer that sits in front of your origins. On the edge service we built for a UK theatre group, a single deployment serves 15+ Broadway and venue domains through a chain of 17 sequential request filters: host resolution, per-site configuration, redirects, CSP and X-Frame-Options headers, and maintenance responses. Consolidating routing for every property into one edge service reduced infrastructure complexity by roughly 90%.

KV is Cloudflare's distributed key-value store, readable from every edge location. We put the things that change more often than code in it: redirect rules, per-domain configuration, maintenance flags. Someone updates a KV entry and the redirect is live, with no code deploy and no release window. That single decision takes most URL and SEO migration work off the engineering queue.

Yes. Maintenance is decided at the edge rather than at the origin, so a window can target one region or one venue while every other property keeps serving normally. Requests inside the window get a proper maintenance response with a Retry-After header, so clients and crawlers back off and return instead of treating the window as a hard failure.

Environment-aware routing sends www, stage, uat, and beta to the right origins from one codebase, and ephemeral preview deployments give a branch its own URL for QA. Staging sits behind IP-gated access and protected environments behind basic auth, so nothing pre-release is publicly reachable.

Yes, and it is one of the better reasons to own the edge layer. Apple Pay domain verification, Android Asset Links, and Apple App Site Association all require exact files served from exact paths on every domain. The Worker serves them centrally for every property, so adding a domain does not mean another origin deploy and another set of files to keep in sync.

TypeScript throughout, Vitest covering the filter chain, and Wrangler-based CI/CD with per-environment configuration. Sentry handles error tracking with tagged scopes, so a failure can be traced to the domain, environment, and filter that produced it. The Worker, the pipeline, and the configuration live in your Cloudflare account and your repository.