How to achieve efficient network information dissemination

   The first point is from the perspective of generating information , That is, the amount of information should be small enough . for instance , from HTTP 1.1 Upgrade to HTTP2 after , stay HTTP 2 Chinese vs HTTP 1.1 Medium REST A large number of... Are transmitted repeatedly in the architecture HTTP header Field , I will use Huffman coding, which is a common compression algorithm . The essence of Huffman coding is to consider the concept of information entropy , Its core principle is to use shorter encoding for repeated words . such as HTTP 2 It's just 5 Bit to write an eight bit single byte number . For those who are very rare, use long codes . In terms of reducing the amount of information , In addition to using Huffman coding , You can also use static tables and dynamic tables .

   Static tables are very common for content , For example, the request method is GET、 The response code is 200 And so on Header Information . Hard code them into the middleware code of the client server , For example, it appears i Express GET request , The presence of a number represents POST request . If hard coded into code , It can be well parsed . Static table volume is finite , The most effective method is dynamic table .

   Before explaining the dynamic table , First, let's explain the relevance of the time dimension . For example, compress high-definition pictures , If you compress it tenfold , Basically, there is no distortion , But it has reached its limit . But if the HD video is compressed , It can be compressed to 100 Times to 1000 times , But it may not be obvious yet . This is due to the correlation in the time dimension . For example, video per second is 30 Frame or 60 frame , The changes between these frames and their corresponding pictures are very small . So as long as the first frame is designed as a key frame , So as to completely store this picture . The latter frame is only a change increment relative to the first frame , The amount of information in the second and subsequent frames is very small . therefore HTTP2 The dynamic table in is based on the same principle . The first time information is transmitted ,UA May be Chorme Or is it FireFox, But a page may contain hundreds of resources , It may even be repeated for hours , At this point, you can pair this at both ends UA Do dynamic coding , For example, encoding by integer numbers . All you need to do is transfer integer numbers .