To understand the mechanics of HTTP / 3, you first need to understand how QUIC, UDP, and HTTP / 2 work , as HTTP / 3 is basically an amalgamation of these components. The very name HTTP-over-QUIC already indicates that the data transfer is not done by TCP, but by UDP..
HTTP / 2 uses TCP as the most widespread transfer protocol on the internet. TCP establishes connections through multi-phase handshakes and transfers data packets chronologically; furthermore, a transfer only proceeds if a packet has been transmitted successfully. The security of transfers is given through so-called acks , that is, request and delivery confirmations and verification numbers . The data transferred by TCP contains a header with parameters that help the sender's processes to connect with the receiver's peer processes .
When it comes to complete data transfer, TCP is very reliable. However, the same cannot be said for data congestion. and loading times, because when a data packet is lost , all transfers stop until it arrives. With HTTP / 2, the Internet protocol family reached its limit, because it was not possible to speed up data transfer without a new protocol.
Therefore, Google took the initiative and developed QUIC , its own transfer protocol. QUIC avoids TCP load congestion by resorting to UDP transfer, which is based on datagrams and requires no connection. Like TCP, UDP works at the transport layer, but instead forgoes sender and receiver acknowledgments . Each transfer does not have to wait for the previous one, and the roundtrip between client and server is shortened considerably. The IETF recognized the benefits of this new protocol and introduced it in 2018 as HTTP-over-QUIC, the version that replaced HTTP / 2.
Basically the HTTP transport protocol remains the same. It is still composed of a header ( header ) and a message body ( body ), and uses verbs, cookies and cache. The difference lies in the way the data is transferred and the availability of built-in encryption.