What Is Intermediate System to Intermediate System in Networking?

Think of a massive post office network that needs to deliver letters across an entire country. To do this efficiently, the post offices need to know how to route letters from one city to another, using highways, local roads, and bridges. The post offices send each other updates about which routes are open or closed, which highways are faster, and which roads are under construction. This way, when a letter arrives, the post office can decide the best way to get it to its destination.

Intermediate System to Intermediate System, or IS-IS, works exactly like this. It is a language that routers use to exchange maps of the network. Each router tells its neighbors about all the destinations it knows about and how far away they are. Over time, every router builds a complete map of the entire network, much like every post office would have a full map of roads and sorting centers.

IS-IS is different from some other routing protocols because it does not use IP addresses to communicate between routers. Instead, it uses its own addressing system, which makes it very flexible and secure. It was originally designed for a network architecture called OSI, but it was adapted to work with TCP/IP networks. Today, it is widely used in large internet service provider backbones, data center networks, and some large corporate networks because it can handle thousands of routers and thousands of routes without losing performance.

When a router receives a packet, it looks at the destination address, checks its map, and finds the best path. If a link breaks or a new router is added, IS-IS quickly recalculates the paths and updates all routers in the area. This makes the network resilient and self-healing.

IS-IS is a link-state routing protocol that operates at Layer 2 of the OSI model, the data link layer. It was originally defined in ISO 10589 for routing in OSI Connectionless Network Service environments, but it was later extended to support IP routing via RFC 1195, which added support for TCP/IP. Unlike OSPF, which is an IP-based routing protocol, IS-IS runs directly over the data link layer using its own packet format, making it protocol-independent and able to route both IP and CLNP traffic.

IS-IS uses a shortest path first algorithm, specifically Dijkstra’s algorithm, to compute the best routes. Each router, called an Intermediate System in OSI terminology, builds a link-state database that contains a complete map of the network topology. Routers flood link-state packets to all other routers in the area, ensuring that every router has the same view of the network. This flooding process is reliable and uses sequence numbers to prevent loops.

IS-IS divides the network into areas. There are Level 1 routers that know the topology within their area, Level 2 routers that know the topology between areas, and Level 1/2 routers that act as border routers connecting areas. This hierarchical structure reduces the size of the link-state database and makes the protocol highly scalable. In large service provider networks with thousands of routers, IS-IS is often preferred over OSPF because it handles the two-level hierarchy more efficiently and converges faster after a topology change.

IS-IS uses a metric called link cost, which defaults to 10 on all interfaces unless manually configured. The path cost to a destination is the sum of the costs of all links along the path. The router selects the path with the lowest total cost. The protocol supports variable-length subnet masking, authentication for security, and fast convergence using features like incremental SPF and prefix independent convergence.

In modern data centers, IS-IS is also used as the routing protocol for fabric networks, such as in Cisco’s Application Centric Infrastructure or in Juniper’s data center designs. It is known for being very stable, with proven performance in networks that require five nines availability.

IS-IS matters because it is the routing protocol of choice for many of the largest networks in the world, including tier-1 ISPs, large data centers, and backbone networks. When you use the internet, your traffic is likely passing through routers that run IS-IS at some point. Understanding IS-IS helps network engineers design, troubleshoot, and optimize these critical infrastructures.

One key reason IS-IS is so important is its scalability. In networks with thousands of routers and tens of thousands of routes, IS-IS performs better and converges faster than many other protocols. It uses a two-level hierarchy that keeps routing tables smaller and update traffic contained within areas. This means that a change in one area does not affect the entire network, which is essential for uptime and stability.

IS-IS is also very flexible. Because it runs directly on the data link layer, it can route both IPv4 and IPv6 without needing separate configurations for each protocol. This makes it a popular choice for dual-stack networks that are transitioning to IPv6. It also supports traffic engineering extensions, allowing operators to control traffic flows across the network for optimal use of bandwidth.

For professionals working in service provider or large enterprise environments, knowing IS-IS is not just an academic exercise. It is a practical skill needed to configure routers, interpret routing tables, diagnose routing loops, plan network expansions, and ensure business continuity. Even if you do not work with IS-IS directly, understanding it gives you a deeper appreciation for how the internet keeps working even when links fail or traffic spikes.

On multiple-choice exams like Network+, you will see questions such as: Which routing protocol operates at Layer 2 and uses the Dijkstra algorithm? or Which protocol is commonly used in large ISP networks and supports both IPv4 and IPv6? These questions test your ability to recall characteristics and differentiate IS-IS from OSPF, EIGRP, or RIP.

Scenario-based questions are also common. For example: A network engineer is designing a backbone for a service provider with 500 routers. The network must support both IPv4 and IPv6 and converge quickly after a link failure. Which routing protocol is best suited? The correct answer would be IS-IS, because it scales well, supports dual-stack, and is commonly used in ISP backbones.

Troubleshooting questions may present partial output from a show command. For example: You see a list of IS-IS neighbors with state UP but some are listed as Level 2 only. Why would a router be Level 2 only? The answer: The router is configured as a Level 2 router and does not participate in intra-area routing.

Configuration questions require you to recognize the correct syntax. For instance: Which command configures IS-IS on a Cisco interface? The answer could be ip router isis. You may also see questions about the IS-IS area format, which uses a 1-byte area ID followed by a 6-byte system ID.

Architecture questions may ask about IS-IS areas and levels. For example: What is the role of a Level 1/2 router in IS-IS? The answer: It connects Level 1 areas to the Level 2 backbone. Understanding the difference between Level 1 and Level 2 is critical for these questions.

Some advanced questions may involve metric calculation. For instance: If all interfaces have a default metric of 10, what is the total cost to a destination four hops away? The answer is 40. These questions test your understanding of how IS-IS builds the shortest path tree.

IS-IS is a protocol that every network engineer working with service providers or large data centers should know. In practice, you will configure IS-IS on a router by first enabling the protocol globally using a command like router isis, then specifying the NET (Network Entity Title) which uniquely identifies the router. The NET includes the area ID and system ID, for example: 49.0001.1921.6800.1001.00. The area ID 49.0001 tells the router which Level 1 area it belongs to.

Next, you need to enable IS-IS on each interface that should participate. On a Cisco router, the command is ip router isis (or clns router isis for CLNP). You may also set the interface type (point-to-point or broadcast) and the metric. By default, the metric on each interface is 10. You can change it to influence path selection, for example setting a higher metric on slower links to discourage traffic.

Verification is done using commands like show isis neighbors, show isis database, and show ip route isis. The neighbor output shows which routers are adjacent, their state, and the level. The database output shows the link-state database, which you can inspect for missing or corrupt entries. If you notice a missing route, you can check if the neighbor adjacency is down or if the link-state database is not synchronized.

Common issues include mismatched area IDs between Level 1 neighbors, authentication mismatches, or level mismatches where one router expects Level 1 but the other is configured as Level 2. Another issue is the use of different metric styles: narrow metrics (old style, up to 63) versus wide metrics (new style, up to millions). If two routers use different metric styles, they cannot exchange routes properly. The solution is to ensure all routers in the domain use the same metric style.

In production, IS-IS is often tuned for fast convergence. Features like incremental SPF (iSPF), prefix independent convergence, and fast hello timers are used to achieve sub-50 millisecond convergence. Redistribution into IS-IS from other protocols like OSPF or BGP is also common, but careful planning is needed to avoid routing loops.

IS-IS connects to broader IT concepts such as network design, redundancy, and Quality of Service. For example, you can use IS-IS to implement load sharing across multiple equal-cost paths, or you can manipulate metrics to engineer traffic away from congested links. Understanding IS-IS gives you deep insight into how large networks operate, which is a valuable skill for any networking professional.