3.4 IP in Homes and Small Businesses

Follow a client from startup to the Internet

The current course includes IP in homes and small businesses. A small router often combines several functions: Ethernet switching, Wi-Fi access, DHCP, IP routing, address translation, and firewalling. These functions are distinct even when one appliance performs all of them.

A useful troubleshooting question is whether a client has the correct address, a usable local gateway, working name resolution, and an onward/return path. Replacing the router before identifying the failed function can hide the actual cause.

Consider a small office using 192.168.50.0/24. Its router has LAN address 192.168.50.1 and a provider-facing connection. A laptop receives 192.168.50.20/24. The gateway address belongs to the laptop's local subnet; a remote website address does not.

DHCPv4 configuration

Dynamic Host Configuration Protocol (DHCP) can provide a leased IPv4 address and configuration options. A common initial exchange is:

1. Discover: The client seeks DHCP service.
2. Offer: A server proposes an address and options.
3. Request: The client requests the selected allocation.
4. Acknowledgment: The server confirms the lease and parameters.

Initial exchanges can use broadcasts because the client does not yet have normal addressing. DHCP servers and clients use UDP ports 67 and 68, respectively. A relay can carry requests between subnets when a server is not on the client's local link.

Useful options include the subnet mask, router/default gateway, and DNS server. A DHCP server is not automatically the default gateway: one device may allocate addresses while another routes packets.

A lease has a lifetime and renewal behavior. An address successfully obtained yesterday does not prove the current server is reachable or that the assigned options still match the network design.

Make the on-link decision

The laptop uses its prefix to determine whether a destination is on the local link. For 192.168.50.30, it can resolve that host's MAC directly with ARP. For remote 203.0.113.80, it resolves the gateway's MAC, while keeping the remote IP as the packet destination.

The Ethernet destination MAC is the next hop, not necessarily the ultimate IP destination. This is the same forwarding distinction used in larger provider networks.

An incorrect mask can produce subtle symptoms. If the laptop is mistakenly configured with a broad prefix, it may ARP for a destination that should have been sent to the gateway. If the gateway is outside the client's real local subnet, ordinary direct gateway resolution may fail.

DNS is a separate dependency

Domain Name System (DNS) resolves names into relevant resource records, including IPv4 or IPv6 addresses. A browser can fail on a hostname even when the client can reach a test server by IP.

Check the configured resolver, whether the query receives a usable answer, and whether the returned address is reachable. A successful DNS answer does not prove the web server's transport port or application is healthy.

DNS commonly uses UDP or TCP port 53. TCP is not limited to zone transfers, and larger UDP responses are possible with EDNS. Encrypted DNS mechanisms can add other transports; the basic lesson is to distinguish name resolution from IP forwarding.

NAT and port translation

RFC 1918 private addresses are usable internally but are not globally unique public destinations. A small office commonly uses Network Address Port Translation (NAPT), often called PAT, to share an external IPv4 address.

For an illustrative outbound TCP connection:

The translation state maps the reply back to 192.168.50.20:40000. Another client can share the public address using a distinct mapping. These documentation addresses illustrate fields, not an actual Internet connection.

Translation does not fix a missing default route, a broken provider link, or an unreachable destination. It also is not encryption. Stateful firewall policy and address translation are related in many appliances but remain separate functions.

Work through an outage

Suppose local file sharing works but websites do not. Check the laptop's address, mask, gateway, and DNS settings. Then test gateway resolution and the router's provider-facing state. Inspect onward routing, translation state where applicable, and return traffic.

If IP access works while hostnames fail, investigate DNS before changing subnet masks. If small packets work but large transfers stall, consider MTU and ICMP-dependent path discovery. If only one application fails, investigate its transport and service rather than assuming the entire IP network is down.

A client using a self-assigned link-local address may indicate that normal address configuration failed. It does not prove that every physical link is disconnected.

IPv6 contrast

IPv6 clients use Neighbor Discovery instead of ARP. Router Advertisements help identify default routers and can support address configuration; DHCPv6 can provide other supported configuration. Do not assume that DHCPv6 supplies the default gateway in the same way as DHCPv4's router option.

Normal IPv6 addressing usually avoids the address-conservation need for IPv4-style NAPT, but routing and firewall policy still matter. “No NAT required” does not mean “every inbound connection is permitted.”

Primary references

DHCPv4, traditional NAT/NAPT, and IPv6 Neighbor Discovery, checked October 20, 2026.