Glossary

IS-IS

Intermediate System to Intermediate System

What is IS-IS

IS-IS (Intermediate System to Intermediate System) is a link-state dynamic routing protocol originally designed by the International Organization for Standardization (ISO) for its connectionless network protocol CLNP. It operates as an interior gateway protocol (IGP), exchanging routing information between devices within a single autonomous system, and calculates paths using the Shortest Path First (SPF) algorithm. IS-IS runs directly over the data link layer rather than over IP, giving it fast convergence, strong scalability, and solid resistance to certain classes of attack — qualities that have made it a long-standing choice for large-scale service provider and carrier networks.

As TCP/IP became dominant, the IETF extended the original OSI-only protocol in RFC 1195 so it could carry both OSI and IP routing information side by side — a variant known as Integrated IS-IS or Dual IS-IS, which is what's deployed in virtually all IP networks running IS-IS today.

Structurally, IS-IS organizes a network into Level-1 (intra-area) and Level-2 (inter-area, backbone) routing, similar in spirit to OSPF's areas and backbone — but the mechanics differ in a few important ways. In IS-IS, every interface on a given router belongs to the same area, whereas in OSPF different interfaces on one router can sit in different areas. IS-IS also has no single designated "area 0" — instead, Level-2 routers anywhere in the network collectively form the backbone, and both Level-1 and Level-2 routing run their own independent SPF calculation, rather than OSPF's model where SPF applies only within an area and inter-area routes are learned via the backbone.

How IS-IS Works

IS-IS's operation is built around three router roles, a hierarchical link-state database, and a defined set of PDUs (protocol data units) for discovering neighbors and flooding topology information.

· Router roles: A Level-1 router only routes within its own area and forms neighbors with other Level-1 or Level-1-2 routers there; destinations outside the area are simply forwarded toward the nearest Level-1-2 router. A Level-2 router carries inter-area routing and must stay physically contiguous with other Level-2 routers to keep the backbone intact. A Level-1-2 router does both, maintaining separate Level-1 and Level-2 link-state databases (LSDBs) and acting as the bridge between an area and the backbone.

· Network types and DIS election: On broadcast links (like Ethernet), IS-IS elects a Designated Intermediate System (DIS) to represent the segment as a pseudo node and flood its link-state information — the router with the highest configured DIS priority wins, with MAC address as the tiebreaker, and Level-1 and Level-2 DIS are elected independently. Point-to-point links (PPP, HDLC) skip DIS election entirely. Notably, unlike OSPF's DR — which only current non-zero-priority routers may become — IS-IS lets priority-0 routers participate in DIS elections, and a newly added router meeting the criteria immediately takes over as DIS, triggering a fresh flood of LSPs.

· Addressing via NET: Each router is uniquely identified by a NET (Network Entity Title), an NSAP-style address whose Area Address portion must match across all Level-1 routers in the same area, and whose System ID portion (48 bits) uniquely identifies the device — commonly derived from that device's IP address for consistency.

· Neighbor establishment and flooding: Routers exchange Hello PDUs to form adjacencies (moving through Down → Initial → Up states), then use LSPs (Link State PDUs) to flood topology information and CSNP/PSNP (sequence number PDUs) to keep link-state databases synchronized — periodically on broadcast links, and at adjacency setup plus on-demand for point-to-point links.

· Route cost styles: Early IS-IS (ISO 10589) capped interface cost at 63 — the "narrow" style. RFC 3784 introduced "wide" costs up to 16,777,215 per interface, letting route costs scale for much larger topologies. Narrow and wide styles can't interoperate directly, so all routers on a network need a consistent style — or a "transition" setting that sends both simultaneously — to stay compatible.

· Overload signaling: A router can mark itself with the IS-IS overload flag — automatically if it can't reliably maintain its LSDB, or manually ahead of planned maintenance — so other routers stop using it for transit while its own directly connected routes stay reachable.

Why IS-IS is Beneficial

The core value of IS-IS lies in fast, scalable, link-state routing for large or carrier-grade topologies:

· Fast convergence at scale: Because IS-IS runs SPF independently at each level and floods topology changes efficiently via LSPs, it converges quickly even as the network grows.

· Protocol independence: Running directly over the data link layer rather than over IP lets IS-IS carry both OSI and IP routing information side by side, and insulates it from certain IP-layer issues that can affect other IGPs.

· Flexible hierarchy without rigid area boundaries: The Level-1/Level-2 model gives the same intra-area/inter-area separation as OSPF's area design, without requiring every router's interfaces to be split across multiple areas.

· Layered security options: Interface, area, and routing-domain authentication can each be applied independently, with a choice of plaintext or MD5, so authentication can be tuned to where in the hierarchy it's actually needed.

· Graceful degradation under stress: The overload mechanism lets a struggling or soon-to-be-maintained router stay reachable for its own routes while being automatically routed around for transit traffic — protecting the rest of the network from a single device's problems.

At Asteraix

What We Can Do at Asteraix

AsterNOS supports a full IS-IS implementation for building fast-converging, scalable IGP topologies:

· Core process and NET setup: isis enable turns on IS-IS packet handling, router isis <area_tag> [vrf <vrf-name>] creates the IS-IS process, and net <NET> configures the router's NET identifier — with ip router isis <area_tag> then enabling the protocol per interface.

· Level and cost control: is-type {level-1|level-1-2|level-2-only} sets the router's level (Level-1-2 by default), isis circuit-type overrides it per interface, and metric-style {narrow|transition|wide} together with isis metric [level-1|level-2] <value> controls route cost calculation and its numeric range.

· Network type and DIS tuning: isis network point-to-point switches an interface off the broadcast default, and isis priority <value> (0–127) adjusts DIS election priority per interface.

· Timers and LSP behavior: isis hello-interval and isis hello-multiplier tune neighbor detection per interface, while lsp-gen-interval, lsp-mtu, max-lsp-lifetime, lsp-refresh-interval, and spf-interval give control over how often LSPs are generated, how large they can be, how long they live, and how frequently SPF recalculates.

· Authentication at every layer: isis password {clear|md5} <password> secures interface-level Hello exchanges, area-password secures Level-1 CSNP/PSNP/LSP traffic, and domain-password does the same for Level-2 — each independently selectable between plaintext and MD5.

· Route origination and redistribution: default-information originate {ipv4|ipv6} {level-1|level-2} always injects a default route into IS-IS, and redistribute {ipv4|ipv6} {ospf|kernel|connected|static|rip|bgp} {level-1|level-2} brings routes from other protocols into the IS-IS domain, both with optional metric and route-map control.

· Fast failure detection: isis bfd links an interface's IS-IS adjacency to BFD, with isis bfd profile <profile-name> available to bind a custom BFD timing profile in place of the 3×300ms default.

· Display and maintenance: show isis neighbor [detail] confirms adjacency state per neighbor, show isis interface reports per-interface IS-IS status, show isis route [level-1|level-2] shows the computed routing table, and show isis database inspects the link-state database directly.

· Typical deployment: Three switches connect two VMs across separate subnets, all running IS-IS. After bringing up interface and Loopback0 IPs, each switch is configured with its IS-IS level and NET, interface-level MD5 authentication is applied on the links between switches, and connected routes are redistributed into IS-IS on the two edge switches so their local subnets propagate across the topology. show isis neighbor detail confirms each adjacency reaches the Up state (with DIS status shown per link), and show ip route isis confirms both edge switches have learned each other's directly connected subnet — verified end-to-end with a ping between the two VMs.