Lesson 3: IPv4, IPv6, Gateways, and Addressing Modes

An IP address by itself does not tell you whether a computer is configured correctly.

The useful question is whether the address, prefix, gateway, addressing method, and IP version make sense together for the network the endpoint is supposed to use.

A typical IPv4 configuration might look like:

address: 192.168.40.25
prefix:  /24
gateway: 192.168.40.1
DNS:     192.168.40.10

Every field answers a different question.

The address identifies the endpoint on an IP network

IPv4 uses 32-bit addresses commonly written as four decimal octets.

Private address ranges include:

10.0.0.0/8
172.16.0.0/12
192.168.0.0/16

These ranges are commonly used inside private networks and are not globally routed as normal public internet addresses.

Public IPv4 addressing follows provider and global routing design.

For support work, the important point is recognizing whether the address is plausible for the network the device is expected to join.

The prefix tells you what counts as local

The subnet mask or prefix length defines which part of the IPv4 address identifies the local network.

With:

192.168.40.25/24

the /24 means the first 24 bits define the network portion in this simplified example.

That lets the host decide whether a destination is local or needs to be sent toward a router.

If two devices have addresses that look similar but use incompatible prefix information, their idea of "local" may not match.

The address and prefix have to be read together.

The default gateway is the exit from the local subnet

The default gateway is normally the router the endpoint uses when the destination is outside its local IP network.

This creates one of the most useful network-fault patterns:

local printer: reachable
same-subnet PC: reachable
remote site: unreachable
gateway: wrong or missing

What does healthy local communication tell you?

The NIC, local link, switching path, and local IP configuration are working well enough for same-subnet communication.

The failure moves toward the gateway or path beyond it.

"No internet" did not mean the whole network was broken.

Static and dynamic describe how configuration is assigned

Static

A static address is intentionally configured and remains until someone changes it.

This can make sense for systems that need predictable addressing, depending on the network design.

Dynamic

Dynamic configuration is commonly supplied through DHCP.

A client requests configuration and receives values such as an address, prefix, gateway, and other options from the DHCP service.

Dynamic does not mean random chaos.

It means configuration is managed through a service rather than typed directly into the endpoint.

APIPA is a clue about failed expected configuration

On Windows, an address in the 169.254.x.x link-local range can appear when the system does not obtain expected IPv4 configuration through DHCP.

Suppose the network expects:

192.168.40.0/24
DHCP enabled
gateway 192.168.40.1

but the endpoint shows:

169.254.33.18/16
gateway: none

What does that suggest?

The interface created a local link address but did not receive the expected site configuration.

That does not prove the NIC is dead.

In fact, the operating system has enough interface function to assign an address to it.

The stronger next question is: why did DHCP configuration fail to arrive?

IPv6 is not just a longer IPv4 address

IPv6 uses 128-bit addresses written in hexadecimal.

It was designed with different addressing behavior and does not simply copy IPv4's private-address model with more digits.

At technician depth, you should be able to recognize:

• that an IPv6 address is present;
• whether the network expects IPv4, IPv6, or both;
• that prefix length still defines network scope;
• that link-local and globally routable IPv6 addresses serve different roles.

Do not disable IPv6 as a ritual because you do not recognize the address format.

Read the actual network design first.

Dual-stack means two IP systems may be working at once

Many networks support both IPv4 and IPv6.

That means one protocol path can work while another fails.

For example:

IPv4 remote access: failing
IPv6 remote access: working

That evidence changes the investigation dramatically.

The physical link and application may be healthy. The failure may belong specifically to IPv4 configuration or routing.

Again, the network is not one yes-or-no state.

DNS belongs beside the address, but it solves a different problem

DNS lets applications use names instead of requiring users to remember IP addresses.

The endpoint can have perfectly valid IP connectivity while DNS is wrong.

That means:

ping known IP: works
open site by hostname: fails

can be consistent with healthy IP reachability and failed name resolution.

We will go deeper into DNS next week.

For now, keep the jobs separate.

Worked case: APIPA on one endpoint

Evidence:

switch link: up
other devices on same switch: normal DHCP addresses
problem PC: 169.254.22.19/16
problem PC gateway: none

What has been proven?

• the interface reports link;
• the switch and DHCP environment work for other endpoints;
• the problem client did not receive expected DHCP configuration.

Strong next checks may include client DHCP state, VLAN or port configuration when managed, DHCP request path, and local OS network configuration.

Weak response:

Type a random public DNS server and hope.

DNS does not assign the missing IPv4 lease.

Worked case: local works, remote fails

Configuration:

address: 192.168.40.25/24
gateway: 192.168.50.1

Local devices on 192.168.40.0/24 work.

Remote destinations fail.

The gateway is not on the endpoint's expected local subnet.

That is a configuration relationship problem, not proof of a bad cable.

Classify endpoint states

For each supplied configuration, identify:

• IPv4 private, public, or link-local clue;
• IPv6 present or absent;
• static or dynamic expectation;
• prefix plausibility;
• gateway plausibility;
• APIPA clue when present;
• what the current state proves;
• next useful evidence.

Before you move on

An endpoint configuration is not one number.

Read it as a system:

address
+
prefix
+
gateway
+
assignment method
+
IP version
+
DNS when names are required

When one relationship is wrong, the symptom depends on which destinations or services rely on it.

Next we build an entire small network from the provider handoff inward and use this addressing model to make every boundary visible.