SMC Networks 8926EM, Руководство по управлению

Страница: 346 / 792

Unicast Routing
20-2
20
Configuring the Routing Information Protocol
The RIP protocol is the most widely used routing protocol. The RIP protocol uses a
distance-vector-based approach to routing. Routes are determined on the basis of
minimizing the distance vector, or hop count, which serves as a rough estimate of
transmission cost. Each router broadcasts its advertisement every 30 seconds,
together with any updates to its routing table. This allows all routers on the network
to learn consistent tables of next hop links which lead to relevant subnets.
Command Usage
• Just as Layer 2 switches use the Spanning Tree Algorithm to prevent loops, routers
also use methods for preventing loops that would cause endless retransmission of
data traffic. RIP utilizes the following three methods to prevent loops from occurring:
- Split horizon – Never propagate routes back to an interface port from which they
have been acquired.
- Poison reverse – Propagate routes back to an interface port from which they
have been acquired, but set the distance-vector metrics to infinity. (This provides
faster convergence.)
- Triggered updates – Whenever a route gets changed, broadcast an update
message after waiting for a short random delay, but without waiting for the
periodic cycle.
• RIP-2 is a compatible upgrade to RIP. RIP-2 adds useful capabilities for plain text
authentication, multiple independent RIP domains, variable length subnet masks,
and multicast transmissions for route advertising (RFC 1723).
• There are several serious problems with RIP that you should consider. First of all,
RIP (version 1) has no knowledge of subnets, both RIP versions can take a long
time to converge on a new route after the failure of a link or router during which time
routing loops may occur, and its small hop count limitation of 15 restricts its use to
smaller networks. Moreover, RIP (version 1) wastes valuable network bandwidth
by propagating routing information via broadcasts; it also considers too few
network variables to make the best routing decision.
A
1
3
6
4
2
5
B C
D E
A
A
B
C
D
Link Cost
0
E
1
1
3
1
0
1
2
1
2
Cost = 1 for all links Routing table for node A