Boosting Internet App Quality: How CDNs Work and Why They Matter
This article explains key internet application quality metrics such as service success rate, latency, and playback smoothness, examines common deployment models, and provides a detailed overview of Content Delivery Networks—including their architecture, operation processes, content injection methods, request scheduling, and distribution strategies—to help improve performance and reliability.
1 Internet Application Quality Overview
1.1 Internet Application Quality
Key QoE (Quality of Experience) indicators include:
Service success rate – the probability that a user request completes successfully.
Service establishment time – the time from request to service presentation, varying by request type.
Latency – the time from user request to result.
Audio/video stutter – pauses caused by insufficient data or low frame rate.
Image clarity – the level of detail and color fidelity in pictures or video frames.
1.2 Network Performance – Latency
Full latency consists of terminal processing delay → network delay → server response delay → network delay → terminal response processing delay.
Network delay is the time packets spend traversing one or more network segments, including routing, ADU transmission, server processing, and propagation.
Transmission delay = frame length / transmission rate.
Propagation delay = distance / propagation speed.
Processing delay is the time routers/servers need to examine packet headers and decide forwarding.
Queueing delay is the cumulative waiting time of PDUs in transmission queues.
1.3 Network Performance – QoS
Latency, jitter, bandwidth, and packet loss together form QoS (Quality of Service) metrics, reflecting underlying packet‑level performance.
2 Common Internet Application Deployment
2.1 Centralized Deployment
Centralized deployment suffers from single‑point failures, limited scalability, and potentially high network latency.
2.2 Website Mirror Acceleration
Mirroring copies an entire site or parts of it to other servers, reducing load on the origin and distributing traffic.
However, mirroring can be cumbersome for frequently updated content and is rarely used by commercial sites.
2.3 CDN Acceleration
CDN combines intelligent mirroring, caching, and traffic scheduling. Its main advantages are:
Alleviates origin server load.
Optimizes hotspot distribution and reduces backbone traffic.
Improves user access quality and overall site speed.
Enhances service reliability and handles traffic spikes.
Mitigates inter‑operator connectivity issues.
Increases security by defending against abnormal traffic attacks.
3 CDN Basic Concepts
3.1 Definition of CDN
A Content Delivery Network adds an extra network layer to deliver origin content to edge locations nearest to users, improving response speed.
Content includes static and dynamic assets.
Distribution means delivering user‑requested content to the nearest node via specific strategies.
CDN consists of thousands of distributed servers communicating over telecom operators' broadband networks.
3.2 Basic Principles of CDN
By deploying cache servers worldwide and using global scheduling and content distribution, CDNs place needed content close to users, turning inefficient IP networks into high‑performance, reliable smart networks.
4 CDN Working Process
4.1 Basic Workflow
Content injection – the first step where origin content is pushed into the CDN.
User request scheduling – directing user requests to the optimal CDN node containing the content.
Content distribution – delivering content from the CDN node to the user or pulling it from upstream nodes.
Content service – serving the content from the nearest CDN node to the end user.
4.2 CDN Content Access
Content storage access – content is pre‑injected and permanently stored in the CDN until explicitly deleted.
Content pre‑injection – content is cached but not permanently stored; it may be evicted based on popularity.
Real‑time back‑source – content is fetched from the origin on demand when a user request arrives.
4.3 CDN User Request Scheduling
Global scheduling directs requests to the nearest node based on geographic location.
Local scheduling operates within a region, considering node health and load to assign tasks more precisely.
4.4 CDN Content Distribution
Push method: Proactive distribution from the origin or central repository to edge nodes using protocols such as HTTP or FTP. Typically used for hot content.
Pull method: Reactive distribution triggered by user requests; if the edge node lacks the content, it pulls it from upstream sources. Suited for dispersed access patterns.
Hybrid method: Combines push (pre‑push of popular content) and pull (on‑demand fetching) strategies.
4.5 CDN Content Service
CDN serves both static and dynamic content based on user requests.
(Copyright belongs to the original author; removed if infringing.)
Signed-in readers can open the original source through BestHub's protected redirect.
This article has been distilled and summarized from source material, then republished for learning and reference. If you believe it infringes your rights, please contactand we will review it promptly.
Raymond Ops
Linux ops automation, cloud-native, Kubernetes, SRE, DevOps, Python, Golang and related tech discussions.
How this landed with the community
Was this worth your time?
0 Comments
Thoughtful readers leave field notes, pushback, and hard-won operational detail here.