Posted on Friday, 19th March 2004 by sean
In the IS-IS Intro I presented a few links for the IS-IS protocol. Before I can proceed to the configuration/verification of IS-IS (ie the good stuff), I have to go over the underlying OSI protocols.
IS-IS started life out as the routing protocol for CLNP, or the Connectionless Network Protocol. Integrated IS-IS was later made to route both CLNP and other protocols like IP.
The OSI protocols have two modes of service, connection oriented and connectionless. Don’t confuse these with their IP layer equivalent concepts of TCP and UDP, these services run over layer 3.
Speaking of services, the OSI protocols define both protocols and services at layer 3. For connection oriented uses, Connection Mode Network Service (CMNS) provides the service that builds up and tears down tunnels. Connection Mode Network Protocol (CONP) carries the upper layer data. This mode is of little interest to us, since dynamic, connectionless routing is what we’re after.
Connectionless Network Service (CLNS) provides the network layer service to the transport layer by way of CLNP, the Connectionless Network Protocol. This is most akin to IP.
OSI protocols differ from the TCP/IP protocols in several ways:
- Nodes can belong to only one area, and as such, only have one address. In the IP world, nodes have several addresses
- There is a strict hierarchy in OSI
- Level 0 routing is known as ES-IS “End Station to Intermediate Station”, ie Host to Router.
- Level 1 routing deals with IS to IS communication in the same area
- Level 2 routing deals with inter area communication
- Level 3 routing deals with routing between Autonomous Systems
Level 2 is also called the backbone, quite similar to OSPF’s backbone. Level 3 isn’t of interest here.
The addressing for CLNS/CNLP is also quite different. Addresses can be up to 160 bits (20 bytes) long, and are called Network Service Access Point (NSAP). As we will see, the NSAP also carries information about the layer 4 protocol, just to confuse things.
An example of an NSAP is
Addresses are read from the right. The last byte is the Selector, it signifies the upper layer protocol. The next six bytes are the End System Identifier, usually the MAC address of a host, or the IP address of the loopback on a router.
IS-IS considers all the leftmost bytes to be the area ID. IS-IGRP hacks off the rightmost two bytes (Area in the diagram) for an area, and the rest for the domain.
DECNET and ATM also use this type of addressing.
If the NSEL of the NSAP is zero, then the NSAP refers to the device itself and is called the NET, or Network Entity Title
So, starting from the right
1 byte – NSEL
6 bytes – ESI
2 bytes – Area
Remaining – Domain
In IS-IS, the Area is the concatination of Domain and Area from above.
Some restrictions:
- The address must have an even number of nybbles (ie falls on an octet boundary)
- ESIs must be unique within an area (be it 1 or 2)
- For a device to talk in Level 0/1, they must be in the same area
Next up, the differences between OSPF and IS-IS. Then we can finally get into configuration.
No related posts.
Related posts brought to you by Yet Another Related Posts Plugin.
Posted in Routing | Comments (0)
