Once all of the OSPF routers have synchronized link-state databases, each
router is responsible for calculating the SPF tree for each known destination.
This calculation is done using the Dijkstra algorithm. In order to do calcu-
lations, metrics for each link are required.
OSPF uses a metric referred to as cost. A cost is associated with every out-
going interface along an SPF tree. The cost of the entire path is the sum of
costs of the outgoing interfaces along the path. Since cost is an arbitrary
value as defined in RFC 2338, Cisco had to implement its own method of
calculating the cost for each OSPF-enabled interface. Cisco uses a simple
equation of 10
. The bandwidth is the configured bandwidth
for the interface.
This value may be overridden by using the ip ospf cost command. The
cost is manipulated by changing the value to a number within the range of
1 to 65,535. Since the cost is assigned to each link, the value must be changed
on each interface.
Cisco bases link cost on bandwidth. Other vendors may use other metrics to
calculate the link's cost. When connecting links between routers from differ-
ent vendors, you may have to adjust the cost to match the other router. Both
routers must assign the same cost to the link for OSPF to work.
on-broadcast multi-access networks (e.g., Frame Relay and ATM)
present a special challenge for OSPF. As you know, multi-access networks
use an election process to select a DR and a BDR to represent all OSPF rout-
ers on the network. This election process requires the participation of all
routers on the multi-access network. However, Hello packets are used to
facilitate the communication for the election process. This works fine on
broadcast multi-access because the connected devices on the network can
hear the AllSPFRouters multicast address for the subnet.
Copyright ©2001 SYBEX , Inc., Alameda, CA