IPv4 addressing — machinageist

A host comes up with an address in 169.254.0.0/16 and no default gateway. What does that tell you?

Two hosts on the same segment both fall back to link-local addressing. What can they do?

Nothing — link-local addresses cannot carry traffic
Reach each other, and also reach the internet once one of them advertises itself as a gateway
Reach each other on that segment, and nothing beyond it
Reach each other only after an administrator adds a static route

Why can a host on a private range start a conversation with a public server, while a public host cannot start one with it?

Private addresses are one-way by design and can only be used as a source
The gateway holds a public address and translates on the private network’s behalf, so there is no return path until the inside starts the exchange
ISPs filter inbound traffic to private ranges as a policy choice, and can be asked to stop
Private addresses use a different protocol number that public routers do not recognise

Which of these is not one of the three blocks RFC 1918 reserves for private use?

ping 127.0.0.1 succeeds on a machine whose network cable is unplugged. What has that proved?

A service is bound to 127.0.0.1 rather than 0.0.0.0. What has that achieved?

Traffic to the service is encrypted, because loopback traffic never leaves the host
The service is reachable only from the machine it runs on, which is a real security boundary
The service is reachable from the local network but not from the internet
Nothing meaningful — a firewall rule is the only way to restrict a listening service

Why is 127 missing from the class A first-octet range of 1 to 126?

An organisation in the classful era needed 300 addresses. What made this awkward, and why does the story still matter?

A class C gave 254 usable addresses and a class B gave 65,534, so the choice was between too few and enormous waste — the granularity problem that CIDR fixed
Class B was reserved for governments, so the only lawful option was several class C networks
Nothing was awkward; classful addressing allowed any prefix length between the class boundaries
Routers of the period could not hold more than one class C route, so a second network was impossible

Does any part of a current router’s forwarding decision depend on address classes?

Yes — the class determines the default mask when none is configured
Yes, but only for addresses in the private ranges
No — classful addressing is historical, and it survives because it explains where the default masks and private ranges came from
No, but classes are still used to decide whether an address may be advertised to the internet

Convert the octet 10101000 to decimal.

Why does IPv4 provide roughly 4.3 billion addresses?

Because the four octets are decimal values from 0 to 255, and 255 to the fourth power is about 4.2 billion
Because RFC 791 caps the allocatable space at 2^32 minus the reserved blocks
Because a 32-bit address gives 2^32 distinct values
Because each of the five classes contributes a fixed share that totals 2^32

You run ping against a server and get replies. What have you established about the service you were trying to reach?

The host is reachable — ping uses ICMP and says nothing about whether any service is listening
The service is listening, since a reply requires an open port
The service is listening on TCP but may be refusing connections at the application level
Nothing at all, because ICMP replies are generated by intermediate routers rather than the host