How Does a CDN Supercharge Your Website’s Speed? A Deep Dive into Architecture and Scheduling

Preface

CDN (Content Delivery Network) is often used in front‑end resource handling such as images, video, audio, and HTML. When performance problems arise, CDN is the first solution to accelerate resource delivery. This article explores CDN principles while prompting readers to consider several questions.

Why is network acceleration needed?

What is a CDN?

How does a CDN achieve acceleration?

What security risks does a CDN have?

Why Network Acceleration?

The main goal is to improve network performance and user experience, preventing user loss and achieving business objectives. Network congestion occurs because users must traverse multiple operators and long physical distances, leading to high latency, limited bandwidth, and routing delays.

Users in different regions cross operators and oceans to reach the origin server.

The Internet is logically a large mesh but physically composed of many smaller networks, causing inter‑network connectivity issues.

Inter‑network links are few, creating “single‑point bridges” that become congested.

Routers and gateways introduce processing delays at each hop.

Using only HTTP, most websites suffer slow access and poor user experience.

Geographical distance (e.g., Beijing to San Francisco) multiplies latency due to routing and operator constraints.

What Is a CDN?

A CDN (Content Delivery Network) distributes content via a globally distributed network of proxy servers, placing website content at the network edge closest to users. This shortens access paths and speeds up resource loading.

Purpose: Enable users to retrieve content from the nearest location, alleviating Internet congestion and improving response speed.

Advantages:

Edge nodes reduce cross‑operator and cross‑region latency.

Most requests are served at edge nodes, offloading the origin server.

Fundamentally, CDN solves two problems:

High latency caused by physical distance between users and data centers.

Data‑intensive applications that suffer from packet loss, congestion, and multiple ISP hops.

Core Technologies:

Intelligent virtual network built on existing infrastructure, using edge servers, load balancing, content distribution, and scheduling to deliver content near users.

Key techniques include content storage and distribution.

In simple terms, CDN caches origin content on edge nodes so users fetch the nearest copy.

CDN Access Process

User enters a domain name; the OS queries the local DNS for the IP address.

Local DNS queries the root DNS for the domain’s authoritative server.

Root DNS returns the authoritative DNS record to the local DNS.

Local DNS queries the authoritative DNS for the domain’s IP.

The authoritative DNS returns the CNAME record to the local DNS.

Local DNS queries the smart scheduling DNS for the final IP.

Smart scheduling DNS selects the optimal CDN node IP based on algorithms and policies.

Local DNS returns the selected IP to the user.

User accesses the site using the returned IP.

CDN edge server responds, delivering cached content to the client.

CDN System Architecture

Workflow

Origin server delegates content for a specific DNS domain to the CDN.

Origin publishes content to the distribution system, which pushes or pulls to edge servers.

The distribution system tracks which edge servers cache which content and manages TTLs and leases.

Clients request the appropriate edge server IP via the routing system or Anycast.

Clients pass through a Scrubber server.

Scrubber forwards clean traffic to edge proxies.

Edge proxies serve client requests and report health to the data control system.

If content is unavailable at the edge, the request is routed back to the origin.

Component Overview

Routing System: Directs clients to the nearest or optimal CDN location using DNS load balancing or Anycast.

Scrubber Server: Separates legitimate traffic from malicious traffic to mitigate DDoS attacks.

Edge Proxy Server: Caches and serves content to end users from RAM for fast retrieval.

Content Distribution System: Distributes content to all edge proxies, typically using a tree‑like model.

Origin Server: Hosts the original content that CDN distributes.

Data Control System: Monitors resource usage, latency, downtime, packet loss, and server load, feeding metrics back to the routing system.

Routing Forwarding

Content Distribution

Active Push

Origin server pushes content to edge nodes, allowing users to retrieve the cached copy directly.

Passive Pull

When a user requests content, the edge server checks its cache; if missing, it fetches from the origin and then serves the user.

Scheduling

Scheduling is the core of CDN, handling traffic intake, routing, and selecting appropriate edge nodes to achieve optimal cost, availability, and speed.

Scheduling Types:

DNS Scheduling

HTTP DNS Scheduling

302 Redirection Scheduling

Anycast Routing Scheduling

DNS Scheduling

Based on the local DNS’s outbound IP and ISP, DNS scheduling suffers from TTL cache delays and lack of EDNS support, leading to delayed failover and cross‑region routing issues.

HTTP DNS Scheduling

Clients query a fixed HTTP DNS address to obtain accurate resolution, reducing hijacking risk but increasing request overhead.

302 Scheduling

Uses the client’s IP and a 302 redirect cluster for real‑time traffic steering. Advantages: real‑time scheduling without DNS cache, high accuracy. Disadvantages: additional redirect latency, suitable mainly for large files.

Anycast Routing Scheduling

Leverages BGP Anycast to provide a few external IPs, enabling fast route adjustments and DDoS resistance. Major providers like AWS CloudFront and Cloudflare use this; domestic CDNs often use Unicast due to complexity.

Conclusion

Understanding CDN architecture helps developers use CDN effectively and troubleshoot related issues. The core of CDN performance lies in the scheduling system. For deeper insights into each module’s logic and hardware deployment, refer to more specialized articles.