EIGRP Load Balancing across unequal paths

Unanswered Question

Hi - i have a quick question i hope you can help me with.


2 buildings, 2 links. Both 1G (mini Gbic) fibre handoffs. One is 1G bandwidth, the other is 155Mb.


If i use the bandwidth statement on the interface, will EIGRP do the rest? Or, do i need to employ the varience command?


I saw my cisco rep last week who seem'd think that EIGRP would handle it with the bandwidth statement on the interface, and discouraged me from using Varience.


Any help grately apperciated.


Cheers, James.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Loading.
lgijssel Mon, 10/29/2007 - 11:22
User Badges:
  • Red, 2250 points or more

You should probably take his advice. With a link pair as unequal as this, there will be little to gain from that extra bit of bandwidth.

It would make a very useful backup circuit though.


regards,

Leo

bvsnarayana03 Mon, 10/29/2007 - 11:35
User Badges:
  • Silver, 250 points or more

I agree with Leo. With a pair of circuits having such huge difference, it is a good option to have low capacity as bkup & let high capacity do all the work.

Jon Marshall Mon, 10/29/2007 - 11:24
User Badges:
  • Super Blue, 32500 points or more
  • Hall of Fame,

    Founding Member

  • Cisco Designated VIP,

    2017 LAN, WAN

Hi James


If you make the bandwidths the same on each interface then you do not need variance. EIGRP should then see them as equal-cost paths, bearing in mind this is quite a high level answer as i do not know your topology.


If you want to keep the original bandwidths then you would need variance although i know some of the NetPro's here think variance is not particularly reliable and i have never used it in vengeance.


Be aware that other things rely on the bandwidth statement on the interface eg QOS so you may affect this as well.


HTH


Jon

sundar.palaniappan Mon, 10/29/2007 - 16:26
User Badges:
  • Green, 3000 points or more

James,


EIGRP doesn't use bandwidth alone to calculate the composite metric. It uses a combination of min bandwidth + total delay to calculate the metric.


The best way to determine the appropriate variance value is to look at the metric of the route via the two paths from the topology table and if the metric of the less desirable path falls within 6 times the value of the best metric then the configuration you have noted above would work.


HTH


Sundar

Actions

This Discussion