CDN selects the optimal node to provide accelerated services to end users through scheduling, which is particularly important. In practical applications, four types of scheduling are commonly used, namely, DNS scheduling, HttpDNS, Edge 302 scheduling, and Center 302 scheduling.
The following is a detailed introduction to the four scheduling types:
1. DNS Scheduling
Intelligent computing, providing services to netizens at the nearest best node.
2. HttpDNS Scheduling
The client directly invokes the HttpDNS interface to obtain the cache server IP group, and then selectively sends requests to the cache servers in the IP group, replacing the conventional DNS scheduling policy. This is applicable to the client, and the client needs to modify slightly to make the HttpDNS interface call.
3. Edge 302 Scheduling
For resolution errors caused by user LocalDNS configuration errors, the cache server can determine based on the terminal IP and respond to the terminal’s optimal cache server IP for terminal access through 302 methods. Suitable for large file downloads (not suitable for small file downloads due to an additional 302 hop network delay)
4. Central 302 Scheduling
The scheduling server accepts terminal Http requests and can schedule user requests to the optimal cache server for response through 302 based on the terminal IP and URL request popularity. It is suitable for downloading large files (not suitable for small files due to the network delay of one more 302 hops), and needs to be used in conjunction with consistent hashing and cold/hot separation.
Note: Consistent Hash and Cold/Hot Separation: When multiple nodes in the same region are involved in a consistent hash of files, it is possible to hash cold files to a single node for centralized access, improving the hit rate, and hash hot files to each node for heat sharing. The hot and cold files can be set proportionally.
