CDN Content Deliery Network Content distribution network transfers content from the origin site to the client to accelerate network access speed Static resource caching
CDN Architecture
LocalDNS----CDN DNS CANME
CDN DNS
Get global load balancer IP
Get the local load balancer IP
Find the optimal CDN node IP
The global load balancing service returns the CDN node iP to the user
User requests CDN node to get data
No data is requested to the upper level cache until the server node
content routing
The CDN load balancing system implements the content routing function of CDN.
Distribute user requests to the best nodes in the CDN
Two levels: Global Load Balancer Global Server Load Balance Local Load Balancer Load Balance
Implementation method: DNS system, application layer redirection, transport layer redirection
content distribution technology
The Cache process from source content to CDN nodes, mainstream content distribution implementation methods:
PUSH: Active distribution, initiated by the content management system, distributes content from the source to the Cache node of the CDN.
PULL: Passive distribution technology, driven by user requests, missing in user request content, and obtaining content in real time from the source or other CDN nodes
content storage technology
Source content storage, large storage capacity, distributed storage system
CDN node storage, mainly functions and performance, content format support, high cache hit rate, possibility, stability
Content Management Technology
CDN content management, improve the efficiency of content services, and improve CDN cache utilization
Three aspects, CDN content indexing, CDN local content copying, CDN local content access status information collection
Key Technologies of CDN System Architecture
Collaborative push technology
The new content of the source server is actively pushed to the CDN nodes, and the CDN nodes share with each other, avoiding the cost of copying & updating, and the CDN maintains the mapping table between the CDn nodes and the source content.
Non-collaborative pull technology
The user request is redirected to the nearest CDN node. If the CDN node does not have this content, the content is requested from the origin server and cached
Collaborative Pull Technology
The first step is the same as above. When the CDN node does not have this content, it will request the content from other CDN nodes to avoid directly requesting the source server and reduce the bandwidth of the original server. The technology is not mature.
Why use a CDN
1. Speed up network latency. 2. Relieve the pressure on the original server, 3. Solve the problem of interconnection. 4 Solve the problem of website burst traffic. 5. Reasonable use of Internet resources. 6. Comprehensive prevention of single point of failure. 7Improve user access quality and experience. 8 Provide comprehensive accurate and detailed data statistics. 9 Anti-hacking ddos.
Consistency Guarantee
There is no data consistency problem with active distribution.
Passive fetch cache update is not timely, data consistency problem. Set the cache invalidation time to achieve eventual consistency. Active distribution can be used if the user has high requirements for consistency, or can it be used? The technology of version=xx can also replace the version number with a different way of returning the url for each uploaded image, which is not recommended.
APPs distribution, version number control, the client randomly sleeps for a period of time, making the distribution off-peak. Different channel numbers have different versions of APPs.