Database Synchronization and Routing Traffic Transport


To accomplish synchronization of the IS-IS link-state database, you use distinctive protocol data units referred to as CSNPs and PSNPs. They are often simply known as Sequence Number PDUs or SNPs. The goal of these special packets is to make sure that LSPs are exchanged in a reliable fashion. They do not actually carry the actual LSP data but rather LSP descriptors, or headers, that define the link-state packets.

Complete Sequence Number PDUs

Complete Sequence Number PDU (CSNP) packets contain an address range (a list of the LSPs held by the router), LSP ID, sequence number, checksum, and remaining lifetime value. The DIS multicasts them every 10 seconds so that the receiving ISs can compare the LSP list to their own database and subsequently request any missing LSPs if necessary. CSNPs are also sent out on point-to-point links whenever the connection comes up. Cisco enables you to configure periodic CSNPs on WAN links as well, if desired.

Partial Sequence Number PDU

Partial Sequence Number PDUs (PSNPs) are implemented to be exchanged for acknowledgement purposes to verify that an LSP was received on a LAN or a point-to-point WAN link. PSNPs are also used to request the transmission of the latest LSP. They usually contain only a single LSP descriptor block. Figure 6.4 gives an example of a PSNP in action.

Figure 6.4. Synchronization on a point-to-point link with PSNPs.

Integrated IS-IS Routing Traffic Transport for IP and OSI

So far in this chapter we have focused on the behavior of IS-IS in an OSI-only (CLNS) implementation. Integrated IS-IS, however, has the flexibility to run CLNS (OSI only), IP only, or a dual OSI and IP scenario. The link-state packets carried in an Integrated IS-IS topology can store several variable-length Type, Length, and Value fields (TLV) and there are IP-specific TLV fields to carry IP information. Although it is true that you can use Integrated IS-IS with only IP routing, you must still configure CLNS addresses because the OSI protocols are needed to generate the peer relationships between the routers. ISs still use the aforementioned IS-IS hello mechanisms to establish end system (host) “to “intermediate system (router) adjacencies. A NET address has to be configured to ID the Integrated IS-IS router, allow the SPF algorithm can do its magic, and facilitate Layer 2 forwarding processes.

Regardless of whether a router is running pure IS-IS or Integrated IS-IS, it has to build the CLNS routing/forwarding table to compute the SPF tree to the NETs or OSI addresses of the destination ISs. Cisco IP routers that use Integrated IS-IS as the routing protocol conform to the IP packet handling requirements as specified in RFC 1812 (Requirements for Internet Gateways). IP routers must process the ISO packets relevant to IS-IS operation, as well as other IP router functions such as ICMP and ARP.

The routing information base consists of two databases that are core to the IS-IS operation: the Link-State database and the OSI Forwarding database. The Link-State database is propagated with routing information learned from the update process. The update process produces local link-state information, based on the adjacency database built by the subnetwork-dependent functions mentioned previously. The router then advertises this information to all peers in link-state packets. The router also gets related link-state information from each adjacent neighbor, retains copies of received LSPs, and then re-advertises them to other peers. Routers in the same area will ultimately have identical Level 1 Link-State databases (matching views of the area topology), that are synchronized through the use of SNPs. This synchronization process is mandatory for routing stability within the area.

The Level 2 Link-State database contains area prefix information that combines the areas for Level 2 ( inter-area ) routing.


The Link-State database facilitates the decision making process by generating the Forwarding database, using the shortest path first (SPF) algorithm (also known as the Dijkstra algorithm). The shortest paths to the NETs or OSI addresses of the destination routers is computed by adding the link metrics along the routing path (chain).

On the BSCI exam, you must remember that an IS runs separate SPF processes and link-state databases for Level 1 and Level 2 routing. Therefore, a L1L2 IS runs the SPF algorithm twice and creates separate SPF trees for both levels.


The OSI Forwarding database is the CLNS routing table and consists of only the best IS-IS routes. These routes are used in packet-switching decisions. If a router has multiple sources of routing information, such as static routes and BGP, the Cisco router uses the administrative distance (AD) to prefer one routing source over another. As you know, the protocol with the lowest AD is preferred. The accepted best route is then installed in the routing table.

Intra-area Level 1 routing uses the SysID value in the destination NSAP. L1L2 routers send default routes to Level 1 ISs in their area. If the packet is going to a destination outside of the area, the Level 1 router forwards the packet to the nearest L1L2 router (the nearest exit point based on the best default route to L1L2 routers). Next , the L1L2 IS routes the packet across the Level 2 backbone by using the area ID of the destination area. Remember that Level 2 routing is based on the area ID value in the NSAP. Upon arrival, the packet is picked up by the Level 1 routing process in the destination area and delivered accordingly .

Take the IS-IS network in Figure 6.5, for example. Suppose that Router50A is sending packets to Router51D using Integrated IS-IS. The packets are first default routed to the Area-50 L1L2 router, Router50B, and then into the destination Area-51 at Router51A. Level 1 routing takes over as Router51A sends the packet through the L1L2 backbone router to the destination Router51D. The total cost on this path is 45. However, on the round trip, Router51D routes the packets back through its closest L1L2 IS, Router53B. This router determines that the lowest cost path to Area-50 is via another route altogether ”through Area-53 via Router53A and over to Router50B to the round-trip destination. This is the nature of IS-IS packet switching and is not necessarily an unfavorable scenario for an internetworking environment.

Figure 6.5. The IS-IS routing process can lead to asymmetric results.

As mentioned earlier in this chapter, a grouping of IS-IS is referred to as a domain, similar to an autonomous system. If you are running a pure CLNS network, you can use the Cisco-proprietary ISO-IGRP protocol to actually link multiple domains. ISO-IGRP deciphers the IDI part of the CLNS route to permit interdomain routing. If you are running IP in your environment, then an IP interdomain routing protocol is necessary. Typically, IS-IS is used in conjunction with BGP for routing IP packets. BGP injects external routes, where IS-IS functions as the IGP responsible for internal routing ( mainly next-hop information for the BGP routes). As a result, there is very little overlap in routing information between either source. IS-IS and BGP routes can, therefore, coexist in the routing table.

Realize that IP information does not enter into the SPF equation when you run Integrated IS-IS, even though the IP information is included in the link-state packets. The SPF tree is constructed solely from CLSNS data. IP information is used locally for comparing things such as administrative distance and are tagged as coming through Level 1 or Level 2. This "separation of powers" from the key IS-IS networking mechanism is another factor that makes IS-IS a more scalable solution than OSPF. The SPF tree is unaffected, for instance, if an underlying IP subnet fails and link-state updates are flooded.



Cisco BSCI Exam Cram 2 (Exam Cram 642-801)
CCNP BSCI Exam Cram 2 (Exam Cram 642-801)
ISBN: 0789730170
EAN: 2147483647
Year: 2003
Pages: 170

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net