IS-IS (Intermediate System to Intermediate System): Scalable Link-State Routing Protocol

IS-IS (Intermediate System to Intermediate System) is a scalable link-state routing protocol for large enterprise and carrier networks, providing shortest-path routing, two-level hierarchy and TLV-based extensibility.

Intermediate System to Intermediate System (commonly abbreviated IS-IS) is a link-state routing protocol used to determine paths for packets inside large enterprise networks and service provider backbones. It was originally developed for routing in OSI networks but was extended to carry IP routing information, and today is a common choice in high-scale environments for directing data across network infrastructures and parts of the Internet.

Overview

IS-IS is a link-state protocol: each router exchanges information about its directly connected links so that every router can build a complete map of the network topology. With that map, routers calculate optimal paths using a shortest-path algorithm. The protocol is designed to scale to very large networks and to be extensible through a flexible Type-Length-Value (TLV) structure that allows additional features to be added without redesigning the core protocol.

Key concepts and operation

Adjacencies and discovery: Routers discover neighbors on a shared medium or point-to-point links and form adjacencies by exchanging Hello messages. On broadcast media, IS-IS elects a Designated Intermediate System (DIS) to coordinate certain activities.
Link State PDUs (LSPs): Each router advertises its local state in LSPs that are flooded throughout the area or routing domain. LSPs include information about connected links, metrics, and reachable prefixes.
Flooding and database synchronization: LSPs are propagated to all routers in the relevant scope and are stored in a link-state database. Sequence numbers and aging are used to ensure consistency and to remove stale information.
Path calculation: Routers run a shortest-path first (SPF) algorithm on the database to compute the best path to each destination. Metrics assigned to interfaces influence path selection.
Two-level hierarchy: IS-IS supports Level-1 (intra-area) and Level-2 (inter-area) routing, and a router can participate in both levels to connect areas. This hierarchical design reduces the size of databases that require SPF computation.

Protocol history and standards

IS-IS originated in the OSI world but was adapted for IP routing. Its core behavior and extensions are specified in IETF standards and related documents; early important specifications described its operation for CLNS and later its integration with IP. Subsequent extensions added support for IPv6, traffic engineering, multi-topology routing, and other features used in modern networks.

Advantages and typical uses

Scalability: suitable for large carrier and enterprise networks.
Extensibility: TLV-based advertisements make it straightforward to introduce new capabilities without changing the base protocol.
Operational flexibility: commonly used as the interior gateway protocol (IGP) inside internet service providers and in networks that use MPLS and traffic-engineering features.

Comparison with other IGPs

IS-IS and OSPF are the two main link-state IGPs. Both provide similar core functionality, but they differ in implementation details, packet encodings, and operational practices. IS-IS runs directly over the data-link layer in its original design and was later extended for IP, which some operators find simplifies protocol extensions. OSPF runs over IP. Choice between them is often based on operator preference, existing operational experience, and ecosystem support rather than fundamental capability differences.

Operational considerations

Deploying IS-IS requires attention to area design, metric planning, and resource sizing for SPF computations. Because IS-IS advertises reachability using network-layer addresses, it is typically integrated with addressing and forwarding strategies used by the operator. Monitoring, tuning of SPF timers, and careful LSP generation practices help keep convergence times predictable in large networks.