| IPv6 Network - Dynamic Routing | | Print | |
IPv6 Network - Dynamic Routing
In a simple two- or three-router setup, using static routes isn’t a problem. However, at some point, static routing becomes unmanageable. Currently four routing protocols work with IPv6:
- • RIPng: the IPv6 version of RIP.
- • OSPFv3: the IPv6 version of OSPF.
- • Integrated IS-IS: the OSI IS-IS routing protocol extended for IPv6.
- • BGP-4 with Multiprotocol Extensions.
The Routing Information Protocol (RIP) is an old and very simple routing protocol for small to medium-sized networks. It basically broadcasts the content of the routing table periodically and incorporates these broadcasts from other routers in its own table. A simple “hop count” makes sure that the most direct route is preferred. RIP suffers from two downsides: it doesn’t work too well in large networks because of all the broadcasts, and it takes very long (several minutes) to detect outages and reroute traffic around the failure. These are fundamental problems caused by the way RIP works, so they’re also present in RIPng.
Open Shortest Path First (OSPF) is a much more advanced routing protocol, which keeps maps of the entire network topology. OSPF sends out “hello” packets to see if neighboring routers are still reachable and to find new neighbors. Apart from that, it only sends out updates when there is a change in the network. In this case, all routers execute the Shortest Path First algorithm, and traffic immediately starts taking the new best path. Its quick reaction to outages and mechanisms to contain routing information to a subset of the network make OSPF suitable for networks of all sizes. Apart from the obsolete version 1, there are currently two versions of OSPF: OSPFv2 (for IPv4) and OSPFv3 (for IPv6). They are completely separate protocols that don’t interact when both are enabled.
Intermediate System to Intermediate System1 (IS-IS) is the OSI CLNS routing protocol for use within a single network or organization. It was later extended to carry IPv4 routing information as well as OSI CLNS routing information. The extended version is referred to as integrated IS-IS. The integrated version is very popular with large Internet Service Providers, because although its basic architecture is very similar to OSPF, its implementation makes it better capable of handling very large networks. In addition to CLNS and IPv4, IS-IS can also carry IPv6 routing information. Until not very long ago, this came with the caveat that every link or subnet that runs IPv4 must also run IPv6, and vice versa. However, there is now “multitopology” support for IS-IS, which removes this limitation when it’s not desired. Because of its OSI background, IS-IS is not for the faint of heart: once it’s up and running, IS-IS isn’t too difficult to manage, but it does require that you to set up CLNS/CLNP addresses and routing and suffer the OSI jargon that comes with it. For small to medium-sized networks, IS-IS provides very few, if any, advantages over OSPF. Unlike the other three routing protocols, the Border Gateway Protocol (BGP) is used between the networks of different organizations. BGP makes it possible for packets to find their way from one ISP to the other. BGP is also used by organizations that connect to two or more ISPs.2 The current version of BGP is version 4. Multiprotocol extensions allow BGP4 to be used for different address families, such as IPv4 multicast or IPv6.
1. “Intermediate System” is OSI-speak for “router.”
2. For more information on BGP, see my book, BGP (O’Reilly & Associates, 2002).
Routing protocol support on Cisco routers depends very much on the model and the software image. Larger models such as the Cisco 7200 support all IPv6 routing protocols. (One of the technical reviewers tells me a Cisco 7200 is no longer considered a “large” model. Obviously he never had to carry one from one end of a big building to another!) Smaller models such as the ancient Cisco 2500 support RIPng and BGP (even though it’s not really powerful enough to do much with BGP), often but not always OSPF (depending on the IOS release train and version), and never IS-IS. The small SOHO/DSL models only support RIPng. A Linux or FreeBSD machine (or even a MacOS one) can be turned into a full-fledged IPv6 router by installing the right software. In the beginning, there was Zebra, which implements RIP, OSPF, and BGP each for both IPv4 and IPv6. The makers of Zebra later started working on a commercial version, sold by IP Infusion. The commercial ZebOS has many more features, including IS-IS, IPv4 and IPv6 multicast routing, MPLS, and VLAN switching. Zebra progress slowed down significantly, and eventually, another group took advantage of the fact that Zebra was released under the GNU Public License and started developing its own version under the
name Quagga. Quagga also supports IS-IS. See http://www.bgpexpert.com/hardsoft.php for a more extensive list of BGP implementations. The rest of the chapter will have examples for Zebra because it’s very similar to Cisco and the configuration commands aren’t subject to change to the degree they are in Quagga. The examples may or may not work with Quagga.
| Users' Comments (0) |
|
No comment posted
