When a packet needs
to be routed, the router looks in its routing table for the next-hop IP address
the packet should take to get to the destination. There may be more than one
matching path, in which case the router will look for the "longest
match". The route that has the longest match - the route with the most
bits in the mask set to "1" - will be the route that is used.
Consider the
following three routes from a fictional Cisco router:
I 172.17.0.0 /24
via 172.1.1.1
O 172.17.0.0 /25
via 173.1.1.1
R 172.17.0.0 /26
via 174.1.1.1
This router has
three possible next-hop IP addresses that it can send packets destined for the
network 172.17.0.0. The masks are of different lengths, meaning that the route
with the longest match (again, the route with the most bits set to
"1") will be used. In this example, the RIP route will be used, since
it has the longest match with a mask of /26. The administrative distances do
not matter.
AD does matter when
the masks are the same length, as shown here:
I 172.17.0.0 /24
via 172.1.1.1
O 172.17.0.0 /24
via 173.1.1.1
R 172.17.0.0 /24
via 174.1.1.1
The longest match
rule always precedes the use of AD, but here there is a three-way tie regarding
the masks. They're all /24 (or 255.255.255.0 in dotted decimal). AD will be
used to break this tie.
As mentioned, AD is
a measurement of a protocol's believability. It is important to keep in mind
that the lowest AD will be preferred. And while the routing table will show you
the ADs of the respective protocols, it's a very good idea to know these ADs
before taking the CCNA or CCNP:
Connected route: 0
Static Route: 1
EIGRP Summary: 5
External BGP: 20
EIGRP (Internal):
90
IGRP: 100
OSPF: 110
ISIS: 115
RIP: 120
EIGRP (External):
170
Unknown: 255 (A
router will not believe a source with an AD of 255, and such routes will not be
placed into the routing table.)
The three protocols
we looked at in the comparison were RIP, IGRP, and OSPF. While your first
instinct may be that the OSPF route would be the most believable, IGRP actually
has a lower AD than the other two and would be the route installed in the
routing table.
Since IGRP does not
support variable-length subnet masking and OSPF does, you may never see this
scenario outside of an exam question. But if you do see it in the exam room or in
a production network, you'll understand how an IGRP route could be preferred
over an OSPF route.
Comments