HTTP/2

What changed in HTTP 2.0

Why HTTP/1 is Slow

The response from the previous request must be received before the next request can be processed

• In HTTP 1.0, a new connection was created for every request sent -> inefficient

• In HTTP 1.1, concepts of persistent connections and HTTP pipelining were introduced

  • Connections could be reused,

  • And multiple requests could be sent to the server in advance

• However, a problem arose in that responses could only be received in the order the requests were sent

  • If the first request had an issue and its response was delayed, the responses of subsequent requests would also be delayed

    -> `Head of Line Blocking`

    • A performance degradation phenomenon that occurs when a head packet is delayed in a packet queue in the Network

Features of HTTP/2

• HTTP header data compression

  • Prevents retransmission of fields that are duplicated from previous headers, saving data

  • Previously headers were Plain Text, but HTTP/2 uses a header compression method called HPACK to improve data transmission efficiency

• Server Push

  • The server can send along certain files that will be needed, such as JS, CSS, and image files that the client did not request, in response to a single HTTP request

• Solved the Head of Line Blocking problem of HTTP 1.x

  • HTTP/2 reduced transmission time by sending multiple files in parallel transmission at once

  • Multiple requests and responses can be sent in parallel over a single TCP connection

    • Multiple parallel streams can exist in a single connection

    • If streams are mixed during transmission, they are reassembled at the receiver using stream numbers

• Stream priority

  • As frames from multiple streams can be multiplexed, the need to specify stream priorities arose

  • The client can use a priority tree to specify stream processing priorities for the server

    • The server sets bandwidth so that higher-priority responses are delivered to the client first