10.4 IS-IS Database Synchronization and DIS

2. Link-State Synchronization Workflow

Database synchronization follows distinct operational patterns depending on whether the physical medium is a point-to-point circuit or a multi-access broadcast LAN:

+---------------------------------------------------------------------------------------------------+
| POINT-TO-POINT LINK DATABASE SYNCHRONIZATION                                                      |
|                                                                                                   |
|   Router A                                                Router B                                |
|      |                                                       |                                    |
|      |-------- 1. Point-to-Point IIH (Initializing) -------->|                                    |
|      |<------- 2. Point-to-Point IIH (Up) -------------------|                                    |
|      |-------- 3. Point-to-Point IIH (Up) ------------------>|  [Adjacency Established]           |
|      |                                                       |                                    |
|      |-------- 4. CSNP (Full LSDB Summary) ----------------->|                                    |
|      |<------- 5. CSNP (Full LSDB Summary) ------------------|                                    |
|      |                                                       |                                    |
|      |-------- 6. PSNP (Requests missing LSP B1) ----------->|                                    |
|      |<------- 7. Full LSP B1 -------------------------------|                                    |
|      |-------- 8. PSNP (Explicit Acknowledgment of B1) ----->|                                    |
+---------------------------------------------------------------------------------------------------+

On broadcast LANs, synchronization is driven asynchronously by the periodic CSNP transmissions of the DIS:

1. The DIS periodically broadcasts a CSNP every 10 seconds to 01:80:C2:00:00:14 (L1) or 01:80:C2:00:00:15 (L2).
2. A newly joined router receives the CSNP and audits its local LSDB.
3. For any missing or newer LSPs advertised in the CSNP, the new router multicasts a PSNP soliciting the full LSP.
4. The DIS (or any other router on the segment that has the requested LSP) responds with the full LSP.
5. If the new router possesses an LSP that is newer than what the DIS advertised in the CSNP, the new router immediately floods its newer LSP onto the LAN, forcing the DIS and all peers to update their databases.

3. Designated Intermediate System (DIS) Mechanics

On a multi-access broadcast segment (such as an Ethernet switch interconnecting multiple routers), requiring every router to establish separate link-state adjacencies and originate individual link connections creates an $N(N-1)/2$ scaling problem. To solve this, IS-IS elects a Designated Intermediate System (DIS) for Level 1 and a separate DIS for Level 2.

+---------------------------------------------------------------------------------------------------+
| THE PSEUDONODE CONCEPT ON BROADCAST NETWORKS                                                      |
|                                                                                                   |
|   Physical LAN Topology:                     Logical Pseudonode Topology:                         |
|                                                                                                   |
|       [Router A]     [Router B]                             [Router A]                            |
|            \             /                                       \                                |
|             ============= (LAN)                                   [Pseudonode]                    |
|            /             \                                   (1921.6800.0001.01)                  |
|       [Router C]     [Router D (DIS)]                            /            \                   |
|                                                             [Router B]    [Router C, D]           |
+---------------------------------------------------------------------------------------------------+

The Pseudonode Concept

The DIS creates and advertises a virtual router called a Pseudonode. Instead of every router advertising connections to every other router on the segment, each router advertises a single connection to the Pseudonode with its configured link cost. The DIS then generates a Pseudonode LSP (with a non-zero Pseudonode ID, e.g., 1921.6800.0001.01-00) listing all routers attached to that LAN segment. This reduces the number of link advertisements in the LSDB from $O(N^2)$ to $O(N)$.

DIS Election Criteria

1. Interface Priority: Values range from 0 to 127. In Junos OS, the default priority is 64.
   • Important Trap: Unlike OSPF where a priority of 0 makes a router ineligible to become DR, in IS-IS a priority of 0 is completely valid and can still become DIS if no other router has a higher priority or MAC address.
2. Tiebreaker: If priority values tie, the router with the highest Subnetwork Point of Attachment (SNPA)—which is the Layer 2 MAC address of the interface—wins the election.

Critical Differences: IS-IS DIS vs. OSPF DR

Why IS-IS Does Not Need a Backup DIS

In OSPF, DROthers form adjacencies only with the DR and BDR. If the DR fails and there were no BDR, all routers would have to start database synchronization from scratch, causing significant network drops. In IS-IS, all routers on the LAN form full adjacencies with one another. Every router already possesses an identical, fully synchronized LSDB. If the active DIS fails:

1. Non-DIS routers miss the DIS holdtime (9 seconds with default Junos timers).
2. The remaining routers immediately elect the new DIS based on highest priority and MAC address.
3. The new DIS originates a new Pseudonode LSP and multicasts a CSNP.
4. Existing adjacencies and databases remain, but pseudonode replacement can still trigger SPF and forwarding convergence.