How the Internet Routes a Packet

Subnet masks, ARP, NAT, longest prefix match, TTL and BGP — no packet ever knows its own route

1 / 22

steps

Act 1 of 8·The Envelope

step 1

the IPv4 header — 20 bytes wrapped around your data

version

IHL

DSCP / ECN

total length

identification

flags

fragment offset

TTL

protocol

header checksum

source address — 32 bits

destination address — 32 bits

Three fields carry the routing story: the destination address, which never changes; the source address, so a reply can come back; and TTL, which every router decrements. Note what is not here — no route, no path, no session. Each packet is forwarded independently, and two packets to the same host can take entirely different paths.

Speednormal

You click a link and your machine has to get bytes to a server it has never spoken to, across networks nobody owns collectively. It does this by wrapping the data in an IP header — twenty bytes of metadata — and handing it to the nearest router with, essentially, a shrug.

Mechanisms

IP header20 bytes · no path

Subnet maskAND, then compare

ARPIP → MAC, local only

L2 vs L3per-hop vs end-to-end

NATrewrite + table

Prefix matchlongest wins

TTL64 → 0, then ICMP

BGPAS path announcements

Active Recall

Question 1 / 3

A packet leaves your laptop for a server on the internet. What are the destination IP and destination MAC set to?

Under the Hood

Loading visualizer…

DNSresolving host

TCPconnecting

TLShandshake

HTTPrequesting

DSAverse

How the Internet Routes a Packet

How the Internet Routes a Packet