TCP/IP

Underneath HTTP and almost everything else sits TCP/IP — the pair of protocols that move bytes across the internet without losing them.

The internet is a packet-switched network: data is chopped into small packets, each routed independently, and reassembled at the far end. Two protocols do most of the work — IP handles addressing and routing, TCP handles reliability.

IP: addressing

Every device has an IP address. IPv4 looks like 192.168.1.10 (four bytes). IPv6 looks like 2001:db8::1 and exists because we ran out of IPv4 addresses. IP gets a packet toward a destination by hopping through routers, but it makes no promises — packets can arrive out of order, duplicated, or not at all.

TCP: reliability

TCP builds a dependable stream on top of unreliable IP. It:

• numbers every byte so the other side can reorder them,
• acknowledges what it received, and retransmits what went missing,
• uses flow control and congestion control so it does not overwhelm the receiver or the network.

The three-way handshake

Before any data flows, TCP sets up a connection:

Client  --- SYN --->        Server
Client  <-- SYN/ACK --      Server
Client  --- ACK --->        Server
        (connection open)

This is why a brand-new connection has a little latency cost — a full round trip before the first byte of real data. (TLS then adds its own handshake on top; see the TLS tutorial.)

Ports

An IP address finds the machine; a port finds the program on it. Some well-known ports:

TCP vs UDP

UDP is the no-promises alternative: fire a packet and hope. It has no handshake and no retransmission, which makes it fast and great for video calls, games, and DNS, where a slightly late packet is worse than a lost one. HTTP/3 actually runs over UDP (via QUIC) and rebuilds reliability itself.