variance...

variance...
no mystery here...
the idea of variance is to take a much less equal route in the eigrp topology table and make it usable in the eigrp route table by manipulating it to be more equal...
that is done by multiplying by a variance value...

the trick is determining that variance value to make the two paths more equal...
can you divide? i thought you could...

first there is this: http://insearchofthecert.blogspot.com/2012/07/variance-eigrp.html

notice he suggests multiplying for the example... that's the router's job... you have to determine what to multiply by...

notice the below topology... it is frame-relay only because i felt like making it frame...


r1 has 2 serial connections to the frame switch... it's topology table shows two exactly equal metrics to the other routers' loopback networks...

r1#sh ip eigrp topo

EIGRP-IPv4 Topology Table for AS(1)/ID(1.1.1.1)

Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,

       r - reply Status, s - sia Status

P 2.2.2.0/24, 2 successors, FD is 2297856

        via 10.1.1.2 (2297856/128256), Serial1/1

        via 10.1.1.2 (2297856/128256), Serial1/2

P 3.3.3.0/24, 2 successors, FD is 2297856

        via 10.1.1.3 (2297856/128256), Serial1/1

        via 10.1.1.3 (2297856/128256), Serial1/2

very nice...

r1#sh ip route eigrp

Gateway of last resort is not set

      2.0.0.0/24 is subnetted, 1 subnets

D        2.2.2.0 [90/2297856] via 10.1.1.2, 00:00:47, Serial1/2

                 [90/2297856] via 10.1.1.2, 00:00:47, Serial1/1

      3.0.0.0/24 is subnetted, 1 subnets

D        3.3.3.0 [90/2297856] via 10.1.1.3, 00:00:47, Serial1/2

                 [90/2297856] via 10.1.1.3, 00:00:47, Serial1/1

and the equal metrics show up as paths for those networks in the route table... the bandwidth is the default for serial links, 1544... slow one of them down to say, s1/2 bandwidth 128, and...

r1#sh ip eigrp topo

EIGRP-IPv4 Topology Table for AS(1)/ID(1.1.1.1)

Codes: P - Passive, A - Active, U - Update, Q - Query, R - Reply,

       r - reply Status, s - sia Status

P 2.2.2.0/24, 1 successors, FD is 2297856

        via 10.1.1.2 (2297856/128256), Serial1/1

        via 10.1.1.2 (20640000/128256), Serial1/2

P 3.3.3.0/24, 1 successors, FD is 2297856

        via 10.1.1.3 (2297856/128256), Serial1/1

        via 10.1.1.3 (20640000/128256), Serial1/2

s1/2 takes a huge hit...

r1#sh ip route eigrp

Gateway of last resort is not set

      2.0.0.0/24 is subnetted, 1 subnets

D        2.2.2.0 [90/2297856] via 10.1.1.2, 00:02:13, Serial1/1

      3.0.0.0/24 is subnetted, 1 subnets

D        3.3.3.0 [90/2297856] via 10.1.1.3, 00:02:13, Serial1/1

and gets pushed out of the route table... we did this artificially to illustrate a point... so put on your imagination hat and pretend that you can't make the adjustment by changing the bandwidth setting on the interface... you have to use the variance command because i said so... the link through s1/2 is hobbled... but
look back at the topology table... to make them near equal you can determine the variance by dividing the large number by the small number... it's about 9...

8.9 really... that's your variance... just add variance 9 to eigrp and it will multiply for you...

r1#sh ip route eigrp

Gateway of last resort is not set

      2.0.0.0/24 is subnetted, 1 subnets

D        2.2.2.0 [90/20640000] via 10.1.1.2, 00:00:35, Serial1/2

                 [90/2297856] via 10.1.1.2, 00:00:35, Serial1/1

      3.0.0.0/24 is subnetted, 1 subnets

D        3.3.3.0 [90/20640000] via 10.1.1.3, 00:00:35, Serial1/2

                 [90/2297856] via 10.1.1.3, 00:00:35, Serial1/1

the routes are back...but notice the metrics are the same as before... that would be your router at work for you...