RFC1923 - RIPv1 Applicability Statement for Historic Status

Network Working Group J. Halpern
Request for Comments: 1923 Newbridge Networks
Category: Informational S. Bradner
Harvard University
March 1996
RIPv1 Applicability Statement for Historic Status
Status of this Memo
This memo provides information for the Internet community. This memo
does not specify an Internet standard of any kind. Distribution of
this memo is unlimited.
Abstract
RIP Version 1 [RFC-1058] has been declared an historic document.
This Applicability statement provides the supporting motivation for
that declaration. The primary reason, as described below, is the
Classful nature of RIPv1.
1.0 IntrodUCtion
RIP version 1 (RIPv1) (as defined by RFC1058) was one of the first
dynamic routing protocols used in the internet. It was developed as
a technique for passing around network reachability information for
what we now consider relatively simple topologies.
The Internet has changed significantly since RIPv1 was defined,
particularly with the introduction and use of subnets and CIDR.
While RIPv1 is widely used in private networks, it can no longer be
considered applicable for use in the global Internet.
2.0 RIPv1 restrictions
RIPv1 has a number of restrictions and behaviors which restrict its
useability in the global Internet.
2.1 Classfulness
Chief among these is that it is a classful routing protocol. RIP
packets do not carry prefix masks. The prefix length is inferred
from the address. For non-local addresses, the prefix is always the
"natural" (classful) length. (e.g., 24 bits for a "Class C" network
address.) For networks to which a local interface exists, if the
interface is subnetted with some specific mask, then RIPv1 assumes
that the mask used locally is the correct mask to apply for all
subnets of that network.
This has a number of effects.
1) RIPv1 can not be used with variable length subnetting. In the
presence of variable length subnetting it will consistently
misinterpret prefix lengths.
2) RIPv1 is difficult to use with supernetting. All CIDR supernets
must be eXPloded and advertised to RIPv1 as individual "natural"
classful advertisements.
3) Even when the networks running RIPv1 are themselves only subnetted
in fixed ways, if the remainder of the network has variable
subnetting then one must carefully make sure that RIPv1 does not
destroy the mask information when it passes through those subnets
running RIPv1. Put another way, co-existence with mutual
information exchange between RIPv1 and more advanced routing
protocols is problematic at best. Note that this applies even when
the other routing protocol is RIPv2.
4) The Internet will soon be making use of addresses which appear to
RIPv1 to be parts of Class A networks. Networks using RIPv1 may not
be able to reach all sites assigned the subsections of a single A.
2.2 Simple Distance Vector
RIPv1 is a simple distance vector protocol. It has been enhanced
with various techniques, including Split Horizon and Poison Reverse
in order to enable it to perform better in somewhat complicated
networks.
However, being a simple distance vector protocol, it will run into
difficulty. First and foremost, it will occasionally have to count to
infinity in order to purge bad routes. This delays the convergence
of routing. In order to keep this short, RIPv1 defines infinity as
16 hops. That means that networks with diameters larger than that
can not use RIP. Even getting close to that limit can cause
confusion for some implementations.
3.0 Conclusion
The recommendation of this Applicability statement is that if there
is reason to run RIP in a network environment, one should use RIPv2
(RFC1723).
RIPv1 itself should only be used in simple topologies, with simple
reachability. It may be used by any site which uses fixed subnetting
internally, and either uses a default route to deal with external
traffic or is not connected to the global Internet or to other
organizations.
RIPv1 may also be used as a local advertising technology if the
information to be used fits within its capabilities.
4.0 Security Considerations
RIPv1 includes no security functions. RIPv2 includes a mechanism for
authenticating the sender of the routing information. Sites which
are worried about the vulnerability of their routing infrastructure
and which feel they must run a RIP-like protocol should use RIPv2.
5.0 Authors" Addresses
Joel M. Halpern
Newbridge Networks Inc.
593 Herndon Parkway Herndon,
VA 22070-5241
Phone: +1 703 708 5954
EMail: jhalpern@newbridge.com
Scott Bradner
Harvard University
1350 Mass Ave, Rm 813
Cambridge MA 02138
Phone: +1 617 495 3864
EMail: sob@harvard.edu