Back to posts

Detailed Explanation and Problem Analysis of CDN Technology

6-minute read
CDNWeb Development
Edit

Table of Contents

Different Usages of CDN

Multiple Caches for the Same Resource

  • Description: Implements multiple caches for the same resource via Vary: Accept-Encoding.
  • Implementation Principle:
    • The origin server returns the Vary: Accept-Encoding response header, which the CDN recognizes.
    • When a client makes a request, if it includes Accept-Encoding: gzip, the CDN returns the Gzip compressed version; if the client does not support Gzip, the regular version is returned.

Auto Polyfill

  • Description: Provides different Polyfills for different browser versions.
  • Implementation Principle:
    • Uses custom headers (like a User-Agent set) to differentiate browser versions.
    • The origin server returns a Vary header, and the CDN provides the corresponding Polyfill file based on the client’s User-Agent.

Auto WebP/AVIF

  • Description: Dynamically returns WebP or AVIF format based on the client’s supported image formats.
  • Implementation Principle:
    • The client request header includes Accept: image/avif, and the CDN automatically returns the AVIF image format.
    • If the client does not support AVIF, the default format (like JPEG or PNG) is returned.

Dynamic Acceleration

  • Description: Improves dynamic request performance through dedicated network lines or CDN route optimization.
  • Implementation Principle:
    • Route Optimization: By changing the BGP routing path and combining multi-path detection technology, the optimal link is selected (based on metrics like packet loss rate, RTT, and overall download time).
    • IP Acquisition:
      • HTTP Request: The CDN marks the user’s IP using X-Forwarded-For or ORIG_CLIENT_IP.
      • HTTPS Request: Since the encrypted message cannot be decrypted, the CDN places the user’s IP in the TCP Option to avoid decrypting the HTTPS message. Here’s a specific example: If a user requests an HTML page and the CDN is configured for dynamic acceleration for origin-pull, it won’t go through the public internet but through the CDN’s optimized dedicated line, making the transmission time faster.

Prefetch Acceleration

  • Description: Pre-loads static resources from the origin server to the CDN edge nodes.
  • Implementation Principle:
    • During DNS resolution, the authoritative DNS returns a CNAME record pointing to the CDN’s global scheduler.
    • The CDN fetches the HTML file from the origin server, parses the image URLs, and caches them on the edge nodes in advance.

CDN Scenario Problems

  1. HTML request timeout due to CDN node failure:
    • Possible cause: Node is down or network anomaly.
    • Solution: Switch to another available node or pull from the origin.
  2. Newly added node has 100% packet loss:
    • Possible cause: Incorrect network configuration or hardware failure.
    • Solution: Check network equipment and configuration to ensure the link is normal.

Dispatching Methods

  1. Cost-based:
    • Description: In situations where some countries have few nodes but high traffic, to reduce costs, traffic can be dispatched to nodes in other countries.
  2. Priority-based:
    • Description: High-priority users have higher dispatching accuracy to ensure a quality user experience.

Specifying Cache Time

  • Implementation: Specify the CDN cache time via the s-maxage in the response header.
  • Purpose: To prevent HTML from being cached in the browser, which can lead to inaccurate PV calculations.

IVS Architecture

  • LVS (Linux Virtual Server):
    • A Linux-based load balancing solution that can distribute user requests to multiple backend servers, improving system availability and scalability.
  • Full NAT (Full Network Address Translation):
    • Translates the source IP and port number of requests entering the LVS to the LVS’s own IP and port number, then forwards them to the backend real servers.
  • Layer 4 Forwarding (L4 Forwarding):
    • Performs load balancing based on the transport layer (TCP/UDP), offering high efficiency but lower flexibility.
  • RealServer:
    • The server that handles the specific business logic.
  • Linux Kernel TCP Kernel:
    • Handles the TCP protocol stack, parsing TCP packets and extracting relevant information.

CDN Issues

  1. Overwrite deployment causes cache to not take effect:
    • Cause: The cache for the new version of the file was not refreshed in time.
    • Solution: Actively purge the CDN cache.
  2. Using timestamps may lead to cache penetration:
    • Cause: The timestamp is different for each request, causing the cache to be invalidated.
    • Solution: Remove the timestamp or use a fixed version number.

Professional Technology

  • CDN is for network acceleration:
    • Generally, resources are pushed to OSS, and then the CDN cache is refreshed.

Case Study

  • Use CDN or OSS:
    • CDN is more suitable for dynamic acceleration and global distribution, while OSS is better for storing static resources.

Vercel CNAME Optimization

  • CNAME Value: cname-china.vercel-dns.com
  • Features: Vercel has specifically optimized for the Chinese network, improving access speed and stability.

The Role of CDN

  • Definition: A CDN (Content Delivery Network) is a distributed network used to cache and distribute static resources.
  • Functions:
    • Caches static resources on edge nodes closest to the user to reduce latency.
    • Improves resource loading speed and enhances user experience.
    • Reduces the load pressure on the origin server.
  • Applicable Scenarios: Suitable for frequently accessed static resources (such as images, CSS, JavaScript files, etc.).

The Role of the Origin Server

  • Definition: The origin server is the original storage location of resources, typically a server or a cluster of services.
  • Functions:
    • When there is a CDN cache miss, the request will be sent back to the origin server to fetch the resource.
    • The origin server is responsible for providing the latest version of the resources.
  • Characteristics:
    • The origin server is usually the processing center for dynamic content.
    • In static resource distribution, the origin server plays more of a backup role.

The Essence of OSS

  • Definition: OSS (Object Storage Service) is a highly available and reliable cloud storage service.
  • Functions:
    • Stores massive amounts of static resources (such as images, videos, documents, etc.).
    • Provides high-concurrency access capabilities and supports global acceleration.
  • Essence: OSS is essentially a storage service, focusing on data persistence and management.
  • Advantages:
    • High Scalability: Supports storage of massive files.
    • Low Cost: Pay-as-you-go, suitable for large-scale storage needs.
    • High Reliability: Multiple replicas for storage, ensuring data security.

Collaboration Workflow

  1. Prioritize CDN:
    • When a user requests a resource, the system first tries to fetch it from the CDN.
    • If there is a CDN cache hit, the resource is returned directly without accessing the origin server.
  2. Origin-pull to the Origin Server:
    • If there is a CDN cache miss, the request is sent back to the origin server.
  3. Origin Server uses OSS:
    • After receiving the request, the origin server fetches the resource from OSS.
    • OSS acts as the underlying storage service, providing the final storage and management of resources.
  4. Update Cache:
    • After the origin server returns the resource, the CDN will cache it so that subsequent requests can be fetched directly from the CDN.

References