The Edge Network Infrastructure

Discover the global architecture powering ArylHive. Built natively on Cloudflare Workers using V8 Isolates instead of traditional containers, the Edge Network executes routing, authentication, and origin proxying physically close to your users worldwide.

1. Global Anycast Routing

When a user visits an ArylHive hosted project (e.g., project.aryl.app), their request is automatically routed to the absolute closest physical data center via BGP Anycast routing, never going back to a single centralized server.

flowchart TD subgraph Users EU[User in London] NA[User in New York] AS[User in Tokyo] end subgraph Edge Network (300+ Cities) CF_LHR[Cloudflare London\nWorker Execution] CF_EWR[Cloudflare Newark\nWorker Execution] CF_NRT[Cloudflare Tokyo\nWorker Execution] end subgraph Databases (Replicated Edge DB) Turso[Turso LibSQL Cluster] end EU -- Anycast --> CF_LHR NA -- Anycast --> CF_EWR AS -- Anycast --> CF_NRT CF_LHR -. Sub-ms Read .-> Turso CF_EWR -. Sub-ms Read .-> Turso CF_NRT -. Sub-ms Read .-> Turso

2. V8 Isolate Architecture vs Containers

Most modern deployment platforms rely on Docker containers or virtual machines (Node.js/Go backend). While these are powerful, they suffer from "Cold Starts"—taking several seconds to boot an idle container when traffic spikes.

ArylHive bypasses this completely by running directly on V8 Isolates. Multiple independent code environments run on a single runtime process. The result? 0ms cold starts. Context switching takes microseconds, meaning whether your site gets 1 hit a day or 10,000 hits a second, the response time is identical and instantaneous.

graph LR subgraph Traditional Architecture VM1[Heavy Container A\nNode.js] VM2[Heavy Container B\nNode.js] end subgraph ArylHive Isolate Architecture Process[Single V8 Runtime Process] Process --- ISO1(Context A) Process --- ISO2(Context B) Process --- ISO3(Context C) end style VM1 fill:#3b0707,stroke:#ef4444 style VM2 fill:#3b0707,stroke:#ef4444 style Process fill:#0f172a,stroke:#3b82f6

3. The Edge Controller Pipeline

The Edge Controller (the main Worker script) handles multiple distinct layers of execution during its 5ms average processing window before passing the raw data downstream.

Phase 1: DNS & Validation — The request hostname is analyzed against .aryl.app, .aryl.cloud, or Custom Domains.
Phase 2: Database Routing — A highly concurrent request fetches metadata from Turso DB (deploy hash, config) while simultaneously checking Cache integrity.
Phase 3: Auth Gateway — If the deployment has a JWT requirement (for dashboard APIs) or Password/Email gates (for File Sharing), execution pauses to render a React/HTML modal.
Phase 4: Stream Transport — Assuming authentication passes, pure byte streams are opened sequentially to underlying Origin Buckets (Backblaze B2/Drive) and piped directly back to the user without being loaded into memory, keeping RAM usage sub-10MB.

Next: Learn about Caching Strategies →