Chapter 10. Integrated IS-IS
| This chapter covers the following subjects:
When the terms link-state protocol and IP are mentioned together, almost everyone thinks of Open Shortest Path First (OSPF). Some might say, "Oh, yeah, there's also IS-IS, but Idunnomuchaboutit." Only a few will think of Integrated IS-IS as a serious alternative to OSPF. However, those few do exist, and there are networksprimarily ISPs and carriersthat route IP with IS-IS. IS-IS, which stands for Intermediate System to Intermediate System, is the routing protocol for the ISO's Connectionless Network Protocol (CLNP). It is described in ISO 10589.[1] The first production incarnation of the protocol was developed by Digital Equipment Corporation for its DECnet Phase V.
The ISO was working on IS-IS at more or less the same time that the Internet Architecture Board (IAB) was working on OSPF, and there was a proposal that IS-IS be adopted as the routing protocol for TCP/IP in place of OSPF. This proposal was driven by the opinion, popularly held in the late 1980s and early 1990s, that TCP/IP was an interim protocol suite that would eventually be replaced by the OSI suite. Adding impetus to this movement toward OSI were specifications such as the United States' Government Open Systems Interconnection Profile (GOSIP) and the European Procurement Handbook for Open Systems (EPHOS). To support the predicted transition from TCP/IP to OSI, an extension to IS-IS, known as Integrated IS-IS, was proposed.[2] The purpose of Integrated IS-IS, also known as Dual IS-IS, was to provide a single routing protocol with the capabilities of routing both CLNS[3] and IP. The protocol was designed to operate in a pure CLNS environment, a pure IP environment, or a dual CLNS/IP environment.
Saying that battle lines were drawn might be overly dramatic, but at the least strong pro-ISO and pro-OSPF factions developed. It can be enlightening to read and contrast the coverage of OSPF and IS-IS in the well-known books by Christian Huitema,[4] a past chairman of the IAB, and Radia Perlman,[5] the chief designer of IS-IS. In the end, the Internet Engineering Task Force (IETF) adopted OSPF as the recommended IGP. Technical differences certainly influenced the decision, but so did political considerations. ISO standardization is a slow, four-step process that depends upon the voted approval of many committees. The IETF, on the other hand, is much more freewheeling. A statement made in 1992 has been its informal motto: "We reject kings, presidents, and voting; we believe in rough consensus and running code."[6] Letting OSPF evolve through the RFC process made more sense than adopting the more formalized IS-IS.
On the other hand, the very small but very sophisticated IS-IS user community has proven to be an advantage for both IS-IS implementers and users. Agreement on extensions tends to be reached quickly, and because the users are mostly high-end ISPs and carriers, they hold tremendous leverage over their router vendors. Having a carrier-grade IS-IS implementation is a must for any router vendor that wishes to enter the high-performance network market. Despite the political friction, the OSPF Working Group learned from and capitalized on many of the basic mechanisms designed for IS-IS. On the surface, OSPF and IS-IS have many features in common:
Below these similarities, there are distinct differences. This chapter begins by examining those differences. Integrated IS-IS (henceforth referred to simply as IS-IS) is covered only as an IP routing protocol; CLNS is discussed only when it is relevant to using IS-IS to route IP. And as mentioned already, IS-IS is almost exclusively found in service provider networks where reliability and scalability are paramount; so an additional focus of this chapter is to look at the features of IS-IS that meet these requirements in very large networks. |
EAN: 2147483647
Pages: 233