ISIS

Intermediate System to Intermediate System, commonly abbreviated as IS-IS, is a widely deployed link-state routing protocol used to determine optimal paths for data transmission within large computer networks. Developed originally for the Open Systems Interconnection (OSI) protocol suite, it has since evolved to support Internet Protocol (IP) routing and is now regarded as a foundational element in many service-provider backbone environments.
IS-IS operates by distributing topology information throughout a network and enabling each router to compute the most efficient forwarding paths using a shortest-path algorithm. Its scalability, stability, and flexibility have led to its adoption in modern high-performance networks where reliable, dynamic routing is essential.

Background and Development

The IS-IS protocol emerged in the late 1980s through work carried out at Digital Equipment Corporation, where it formed part of the DECnet Phase V architecture. Its primary function was to support the Connectionless-mode Network Service (CLNS) within the OSI framework.
In 1990, a draft version of the protocol was published by the Internet Engineering Task Force (IETF), but the document was later withdrawn due to conflicts with the International Organization for Standardization (ISO) standard. ISO subsequently standardised IS-IS in 1992 under ISO 10589, defining communication procedures for devices referred to as Intermediate Systems.
Although developed alongside the Internet Engineering Task Force’s Open Shortest Path First (OSPF) protocol, IS-IS later expanded to support IP routing. This extended form, referred to as Integrated IS-IS, enabled seamless operation within mixed OSI and IP environments, increasing the protocol’s relevance as the global Internet evolved.
By 2005, IS-IS had become widely recognised as the de facto choice for large-scale service provider backbones due to its robustness and straightforward scalability.

Core Operating Principles

IS-IS functions as an interior gateway protocol, operating within a single administrative domain. It contrasts with exterior gateway protocols such as the Border Gateway Protocol (BGP), which connect separate autonomous systems.
As a link-state routing protocol, IS-IS depends on the reliable flooding of topology information. Each router maintains a database representing the full network topology and uses Dijkstra’s algorithm to compute the shortest path to every destination. Key elements of IS-IS operation include:

• Independent topology databases, allowing each router to compute routes autonomously.
• Flooding of link-state information to maintain consistent and up-to-date topological views.
• Deterministic path computation, ensuring stable and loop-free forwarding behaviour.

Datagrams are subsequently forwarded using the optimal paths calculated from the topology database.

Terminology and Conceptual Framework

IS-IS incorporates a unique terminology derived from ISO standards. Key terms include:

• Intermediate System (IS): Equivalent to a router.
• Designated Intermediate System (DIS): The designated router for a multi-access network segment.
• End System (ES): The equivalent of a host device.
• Circuit: A network link.
• Adjacency: A neighbour relationship between routers.

IS-IS supports two circuit types:

• Broadcast LAN, used on shared media.
• Point-to-point, used on direct links.

Unlike OSPF, it does not support point-to-multipoint topologies.

Protocol Data Units and Database Synchronisation

IS-IS routers exchange a series of Protocol Data Units (PDUs) to maintain topology information and synchronise databases.
Hello PDUs (IIH): Used to establish and maintain adjacencies. On broadcast networks, separate Level-1 and Level-2 Hello messages are used, whereas point-to-point links rely on a single type. Unlike OSPF, Hello timers do not need to match exactly for adjacencies to form.
Link State PDUs (LSPs): Contain routing information encoded in Type-Length-Value (TLV) structures. Each LSP bears an identifier composed of a System ID, Pseudonode ID, and Fragment ID. LSP fragmentation occurs when large PDUs exceed size limits, with fragments tracked using the Fragment ID field.
Complete Sequence Number PDUs (CSNPs): Sent by the DIS at regular intervals—typically every ten seconds—these list all LSPs known to the DIS, including their sequence numbers and checksums.
Partial Sequence Number PDUs (PSNPs): Used by routers to request missing or outdated LSPs when discrepancies are detected during CSNP comparison.

Network Entity Title (NET) and Addressing

IS-IS diverges from traditional IP addressing by using a Network Entity Title (NET) rather than assigning a Layer-3 address to every interface. A NET serves as the unique identifier for an IS-IS router and is assigned to the loopback interface, which remains consistently operational.
A NET comprises:

• Area Identifier (variable length, consisting of the Address Family Identifier and Area ID).
• System ID (6 bytes).
• NSEL (1 byte), which must be zero for adjacency formation.

Routers may hold up to three NETs, enabling seamless migration between areas.

Hostname Visibility Through TLVs

To simplify troubleshooting and topology inspection, IS-IS includes TLV 137, which carries the originating router’s hostname within every LSP. This enables all routers to associate System IDs with readable hostnames, facilitating easier diagnosis and analysis compared with OSPF, which typically relies on IP addresses.

Areas, Levels, and Routing Behaviour

IS-IS defines a two-level hierarchy:

• Level-1 (L1): Similar to a stub area in OSPF; routers maintain local area reachability.
• Level-2 (L2): Acts as the backbone; routers form a contiguous routing domain.
• Level-1-2 (L1L2): Routers that participate in both levels and function similarly to OSPF ABRs.

Several distinctions characterise IS-IS area operation:

• Areas terminate on routers rather than links.
• L1L2 routers set the Attached (ATT) bit in their L1 LSPs when connected to another area, prompting L1 routers to install a default route toward the L1L2 router.
• L1 routers may inject external routes into their area, with export behaviour to L2 controlled by policy rather than mandatory design constraints.
• Modifying route export policy does not require adjacency resets, enabling more seamless network adjustments than comparable changes in OSPF.

Attribute Bits and Topology Functions

The Attribute block of an LSP contains several control bits:

• P bit: Denotes support for partition repair, a mechanism largely unused in contemporary networks.
• ATT bits: Four bits historically associated with different routing metrics. Their primary modern use is signalling that a router is attached to another area, prompting L1 routers to install a default route.

These bits illustrate the protocol’s original intention to support multiple routing topologies with distinct metric calculations, a concept later streamlined in practice.

Importance in Modern Networking

IS-IS remains a cornerstone of large-scale IP and multi-protocol networks. Its reliance on TLV encoding allows extensibility, enabling new features—such as support for IPv6, multi-topology routing, and traffic engineering—to be added without redesigning the protocol. Its ability to operate efficiently in dense topologies with extensive adjacencies has solidified its role in service provider, academic, and enterprise backbones.