Network interconnection using network layer protocols

Search Transfer and Protection of Information
Theory
Author
Affiliation

Andrei Biziuk

VSTU

Published

June 19, 2025

1. Introduction: The Challenge of Internetworking

Today, we will discuss one of the most fundamental concepts of modern computing: how we connect disparate, independent networks to create a single, logical, global internetwork. This is the core challenge solved by the Network Layer.

Imagine our university’s network, a local coffee shop’s Wi-Fi, and a corporate network in another country. These are all separate, autonomous networks, likely using different underlying technologies—Ethernet, Wi-Fi, fiber optics. How does a packet from your laptop on campus find its way to a server in that foreign corporate network? It’s not magic; it’s the result of carefully designed network layer protocols.

The primary goal of the Network Layer (Layer 3 of the OSI model) is to provide logical host-to-host addressing and routing. It abstracts away the complexity of the underlying physical and data link layers, creating a unified view of the network.

2. Principles of Network Interconnection

To connect different networks, we need a special device that operates at the network layer: a router. Unlike switches (Layer 2) which forward frames based on MAC addresses within a single network, routers forward packets between different networks based on logical addresses (like IP addresses).

The core principles of network interconnection at this layer are:

  1. Logical Addressing: Every host on the internetwork must have a unique, globally significant address. In the TCP/IP world, this is the IP address (e.g., 192.168.1.101). This address contains information about both the specific host and the network it belongs to.

  2. Packet Forwarding (Routing): Routers make decisions about where to send a packet next. When a router receives a packet, it examines the destination IP address, consults its internal routing table, and forwards the packet to the next-hop router that is closer to the final destination. This process is repeated at each router along the path.

  3. Encapsulation: As a packet traverses different networks, it may need to be encapsulated in different types of Layer 2 frames. For example, a packet might travel over an Ethernet segment, be un-encapsulated and re-encapsulated for a WAN link, and then again for another Ethernet network. The Layer 3 packet itself, however, remains intact throughout its journey. This is a key concept of protocol layering.

  4. Fragmentation and Reassembly: Different network technologies have different limits on the maximum size of a data frame they can carry, known as the Maximum Transmission Unit (MTU). If a router needs to forward a large packet onto a network with a smaller MTU, it must fragment the packet into smaller pieces. The destination host is then responsible for reassembling these fragments back into the original packet.

3. Routing Protocols and Routing Functions

A router’s primary job is to build and maintain its routing table. How does it learn the best paths to all possible destination networks? It can be done in two ways:

  • Static Routing: A network administrator manually configures the routes in the routing table. This is simple, secure, and efficient for very small, stable networks. However, it is not scalable and cannot adapt to network changes, such as a link failure.
  • Dynamic Routing: Routers use special routing protocols to automatically communicate with each other, share information about the networks they are connected to, and calculate the best paths. This is the foundation of the scalable, resilient internet we use today.

Routing protocols perform several key functions:

  • Discovery: Learn about neighboring routers and the networks they can reach.
  • Path Selection: Apply a routing algorithm (or metric) to determine the “best” path. The metric could be hop count (the number of routers), bandwidth, delay, load, or a combination of factors.
  • Maintenance: Detect when a path is no longer available (e.g., a link goes down) and find an alternative, loop-free path. This is known as convergence.

4. The OSI Hierarchical Routing Model

To manage the immense complexity of a global network, we don’t treat it as one flat entity. Instead, we use a hierarchical structure, as formalized in the OSI routing model. This model organizes the internetwork into a hierarchy of systems and groups.

  • End System (ES): These are the user devices—computers, servers, smartphones. Their job is to run applications, not to forward traffic for other networks. In TCP/IP terminology, these are simply hosts.
  • Intermediate System (IS): These are devices whose primary function is to route packets between networks. In TCP/IP, we call these routers.
  • Area: A collection of End Systems and Intermediate Systems within a common administrative and routing policy domain. Routing within an area is managed cohesively.
  • Domain (or Autonomous System - AS): This is the highest level of the hierarchy. An Autonomous System is a large collection of connected networks and routers, all under a single administrative authority. Your university campus network is one AS. A major Internet Service Provider (like Comcast or AT&T) manages its own, much larger, AS. Each AS is assigned a unique number (ASN).

This hierarchical model is crucial for scalability. A router inside one AS doesn’t need to know the detailed topology of every other AS in the world. It only needs to know the topology within its own area/domain and which “exit” router to use to reach other domains.

5. Intradomain vs. Interdomain Routing

This hierarchical structure leads to two distinct classes of routing protocols:

5.1. Intradomain Routing Protocols (Interior Gateway Protocols - IGP)

  • Purpose: Used for routing within a single Autonomous System.
  • Goal: Find the most optimal path quickly and efficiently. These protocols can be complex because they need detailed knowledge of the internal network topology.
  • Examples:
    • RIP (Routing Information Protocol): A simple, older protocol based on the Distance Vector algorithm. Its metric is hop count. It’s easy to configure but suffers from slow convergence and is not suitable for large networks.
    • OSPF (Open Shortest Path First): A much more sophisticated protocol based on the Link-State algorithm. Each router builds a complete map (or “link-state database”) of the entire AS and then independently runs Dijkstra’s algorithm to calculate the shortest path. OSPF is highly scalable, converges quickly, and supports hierarchical routing through “areas.”
    • EIGRP (Enhanced Interior Gateway Routing Protocol): A proprietary Cisco protocol that combines the best features of Distance Vector and Link-State algorithms. It offers very fast convergence.

5.2. Interdomain Routing Protocols (Exterior Gateway Protocols - EGP)

  • Purpose: Used for routing between different Autonomous Systems.
  • Goal: Not necessarily to find the “fastest” path, but to enforce high-level policy. An AS administrator defines policies like “Don’t send traffic for Competitor-X through my network,” or “Prefer this expensive, high-quality link only for high-priority traffic.”
  • Example:
    • BGP (Border Gateway Protocol): This is the one and only EGP used on the global Internet. BGP is a Path Vector protocol. It doesn’t just advertise reachability; it advertises the entire path of Autonomous Systems that a packet would take to reach a destination. This allows BGP routers to detect routing loops and, most importantly, to apply complex routing policies based on the path attributes. BGP is the “glue” that holds the entire Internet together.

Think of it this way: IGPs are like the local road maps you use to navigate within a city (an AS), while BGP is the highway and airline system you use to travel between cities (different ASes).

6. Internetworking Based on Protocol Stacks

Finally, let’s tie this back to the protocol stacks we’ve studied. The most dominant protocol stack is TCP/IP.

Here’s how it enables internetworking:

  • Network Layer (IP - Internet Protocol): This is the workhorse. The IP protocol defines the logical addressing scheme (IPv4/IPv6) and the packet format (the IP datagram). It is a connectionless and unreliable protocol, meaning it provides a “best-effort” delivery service. It doesn’t guarantee delivery, order, or error-free transmission. Those guarantees are the responsibility of higher layers, like TCP.
  • Supporting Protocols:
    • ICMP (Internet Control Message Protocol): Works alongside IP to provide error reporting and diagnostics. When you use the ping command, you are using ICMP. If a router cannot deliver a packet, it might send an ICMP “Destination Unreachable” message back to the source.
    • ARP (Address Resolution Protocol): A crucial utility protocol that maps a known IP address to an unknown MAC address on a local network segment. A router needs this to create the final Layer 2 frame for delivery to the next hop.

When an application on Host A sends data to Host B in a different network:

  1. The Transport Layer (TCP) segments the data and adds a TCP header.
  2. The Network Layer (IP) takes this TCP segment, encapsulates it in an IP packet, and adds the source and destination IP addresses.
  3. Host A’s IP layer determines that the destination is on a different network and forwards the packet to its default gateway (a local router).
  4. The router receives the packet, strips off the Layer 2 frame, examines the destination IP address, and consults its routing table (built by OSPF, BGP, etc.).
  5. It forwards the packet to the next-hop router. This process repeats across the internet.
  6. Finally, the packet arrives at the destination network’s router, which delivers it to Host B. Host B’s network layer strips the IP header and passes the segment up to the TCP layer for reassembly and delivery to the application.

7. Conclusion

Network interconnection using network layer protocols is a masterpiece of layered, hierarchical design. By separating the concerns of physical transmission, local delivery, and global routing, we can build a system that is both incredibly complex and remarkably robust.

The key takeaways are:

  • The Network Layer provides logical addressing and routing.
  • Routers are the key devices that forward packets between networks.
  • Routing protocols (like OSPF and BGP) dynamically build the routing tables that guide these packets.
  • A hierarchical model of End Systems, Intermediate Systems, Areas, and Autonomous Systems makes the global internet scalable.
  • We use different types of protocols for routing within an AS (IGPs) and between ASes (EGPs), each optimized for different goals—speed versus policy.

Understanding these principles is essential not only for network engineering but also for cybersecurity, cloud computing, and any field that relies on the reliable and efficient transfer of information across the globe.

Back to top