Week 1 · Lesson 15

OSI Model - Network Layer

Layer 3 is where an internetwork becomes possible. It gives every host a global address and every router a map, so a packet can cross dozens of networks it has never seen before and still arrive at exactly the right machine.

CCNA 200-301 Domain 1.0 Network Fundamentals Verified 2026-10-11
Part 1

Layer 3 in the stack

The network layer is Layer 3 of the OSI model. It sits directly above the Data Link layer (Layer 2) and directly below the Transport layer (Layer 4). Each layer answers a different scope of the same problem — moving data:

The Transport layer gives process-to-process service; the network layer gives host-to-host service across the whole path. In the TCP/IP model, Layer 3 is called the Internet layer, and its defining protocol is IP.

Definition — network layer. The network layer (Layer 3) is responsible for delivering data from a source host to a destination host across multiple networks, using logical (IP) addresses and the routers in between.
Part 2

Four jobs of the network layer

Exams expect you to break the network layer into its distinct functions. There are four:

JobWhat it doesWhere it happens
Logical addressingAssign and use Layer 3 addresses (IP) that identify hosts and networks.Hosts and routers
RoutingChoose the best path from source to destination across the internetwork.Routers (control plane)
ForwardingMove each individual packet toward its destination, one hop at a time.Routers (data plane)
FragmentationSplit a packet too large for a link; the destination reassembles it.Routers (IPv4), destination host

Routing and forwarding are easy to confuse. Routing is the big-picture decision — “which way should traffic go?” — and it fills a table. Forwarding is what the router does with every single packet that arrives, using that table.

Part 3

Logical addressing with IP

An IP address is a logical address: it is assigned in software by an administrator or a DHCP server, not burned into the hardware. Because it is logical, it is independent of the underlying link technology — the same IP can travel over Ethernet, Wi-Fi, or fibre without changing. A MAC address, by contrast, is physical and only meaningful on the local link (Layer 2).

IPv4 addresses are 32 bits long; IPv6 addresses are 128 bits. An IP address has two parts: a network portion, which lets routers summarise and forward whole networks, and a host portion, which identifies the specific device inside that network. This two-level, hierarchical structure is what makes global routing scale. If every address were flat and random, routers would need a separate entry for every host on Earth.

Network and host portions · drag the prefix length
192.168.10.37/24

Several different protocols define how Layer 3 works:

ProtocolPurpose
IPv4The original Internet Protocol (RFC 791). Connectionless and best-effort: it makes no guarantee of delivery, ordering, or error recovery. Those are handled by higher layers.
IPv6The successor to IPv4, using 128-bit addresses and a simplified fixed header. Defined in RFC 8200.
ICMPInternet Control Message Protocol (RFC 792). Carries error and diagnostic messages that sit alongside IP. The tools ping and traceroute depend on it.
Routing protocolsRIP, OSPF (and others) run at Layer 3 to exchange reachability information and build routing tables. In-depth coverage is Week 6.
ARP. One more protocol is often mentioned near Layer 3: ARP (Address Resolution Protocol), which finds the MAC address that belongs to a known IPv4 address. It genuinely straddles the Data Link and Network layers, so it is studied with both.
Part 4

Routing and forwarding

Routing is the process of choosing a path from source to destination across an internetwork. Routers keep a routing table (also called a forwarding table) that lists destinations and the next hop for each. Routers learn routes in two ways: they know their directly connected networks automatically, and they learn everything else through routing protocols such as RIP and OSPF (covered in depth in Week 6).

Note. Routing is a whole-path decision, but no single router ever sees the whole path. Each router only decides the next hop, and the packet reaches its destination because every router along the way makes a consistent local decision.

Forwarding is the per-packet action. A router receives a packet on one interface, looks at the destination IP address, finds the matching entry in its forwarding table, and sends the packet out the correct interface toward the next hop. The router needs no knowledge of the full path — only the next hop.

If several entries match, the router uses the most specific match (longest prefix match). If no entry matches, it uses the default route; if there is no default route, the packet is dropped. A dropped packet may trigger an ICMP error message back to the sender.

Longest-prefix match · choose a destination
Destination prefixNext hopInterface

Part 5

Fragmentation, MTU and routers

Every link has a maximum transmission unit (MTU) — the largest payload a frame on that link can carry. If a packet is larger than the MTU of the outgoing link, IPv4 can fragment it: the router splits the packet into smaller fragments, gives each a copy of the IP header (with fields that let the pieces be identified and ordered), and the destination host reassembles them. Reassembly happens at the destination, not at intermediate routers.

IPv6 changed the rules: IPv6 routers do not fragment. The sending host must discover the smallest MTU along the path (Path MTU Discovery) and send packets that already fit.

Exam tip. Two fragmentation facts are commonly tested: (1) in IPv4, routers may fragment and the destination reassembles; (2) in IPv6, routers do not fragment. Fragmentation is a symptom of an MTU problem, not a normal healthy state.

The device that operates at Layer 3 is the router. Each router interface connects to a different network and has its own IP address. A router’s job, step by step:

  1. Receive a frame on an interface.
  2. Remove the Layer 2 header and trailer (de-encapsulate down to the packet).
  3. Read the Layer 3 destination address and consult the forwarding table.
  4. Build a new Layer 2 frame for the next link (re-encapsulate) and send it.

A host uses a default gateway — a router on its own network — to reach any destination that is not local. If the destination is on the same network, the host sends directly; if not, it sends to the gateway.

TTL and ICMP · send a probe
TTL 1
Host A R1 R2 R3 Host B probe

Part 6

Crossing a network boundary

This is the concept that ties Layer 3 to the whole course. When a packet crosses from one network to another, the IP addresses stay the same from source to destination, but the MAC addresses are rewritten at every router hop. The router is exactly the point where the Layer 2 addressing changes because the link changes.

Crossing a network boundary · step through the hops
Host A R1 R2 Host B packet

Hop 1 of 3

Exam tip. If a question shows a packet arriving at a router, remember: the source and destination IP addresses do not change (barring NAT), but the source and destination MAC addresses in the new frame are the router and the next-hop device. MAC changes hop by hop; IP survives the whole journey.

The Layer 3 Protocol Data Unit is the packet. (Older texts also say “datagram”, and IPv6 formally calls it a packet too.) At Layer 3 the IP packet consists of the IP header plus the payload (the transport-layer segment or datagram). Key IPv4 header fields include:

FieldWhat it is for
VersionWhich IP version this is (4 for IPv4, 6 for IPv6).
Source / Destination addressThe logical addresses that routing and forwarding use.
Time To Live (TTL)A counter decremented at each router; when it reaches zero the packet is dropped, preventing loops (IPv6 calls this Hop Limit).
ProtocolIdentifies the payload carried inside (for example TCP, UDP, or ICMP).
Fragmentation fieldsIdentification, flags, and fragment offset, used for IPv4 fragmentation and reassembly.
Remember the PDU ladder. At Layer 4 it is a segment (TCP) or datagram (UDP); at Layer 3 it is a packet; at Layer 2 it is a frame; at Layer 1 it is bits. Only Layer 3 has packets.
Part 7 · Your turn

Review

Check what you have learned with these review questions.

Review · 6 questions

Key terms

Sources: RFC 791 (Internet Protocol); RFC 1122; RFC 8200 (IPv6); RFC 792 (ICMP); IEEE 802.3 and 802.11; ISO/IEC 7498-1; Cisco CCNA 200-301 exam topics.

Sources

Sources & references

  1. Postel, J. RFC 791: Internet Protocol (1981). rfc-editor.org/rfc/rfc791.
  2. Braden, R. RFC 1122: Requirements for Internet Hosts — Communication Layers (1989). rfc-editor.org/rfc/rfc1122.
  3. IEEE. IEEE 802.3 (Ethernet). standards.ieee.org.
  4. IEEE. IEEE 802.11 (Wireless LAN). standards.ieee.org.
  5. ISO/IEC. 7498-1:1994 — Information technology — Open Systems Interconnection — Basic Reference Model. iso.org.
  6. Cisco. CCNA (200-301) Exam Topics, Network Fundamentals. cisco.com.

Target exams: CCNA 200-301 v1.1 · CyberOps Associate 200-201. Last content verification: 2026-10-11.