WAN Design Question

Unanswered Question
Apr 20th, 2009
User Badges:


We are adding a new WAN Site with 2 X Fe connecting to two other sites which are well connected on 1 Ge. How to achieve link loadbalancing and Hardware failover at the same time. The sites will sit in the corners of a triangle.

Lets say the new site is A and the existing sites are B and C with one NW subnet each. When site A sends traffic to B or C it has to be on the direct link connected (which can be achieved using any IGP).

If we use HSRP for Hardware Failover then we end up using a single link to send traffic out from the site and receive traffic on the other link for the directly connected site (which is not good for all apps). Is there a way to achieve both.

Thanks in advance,


  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
Joseph W. Doherty Mon, 04/20/2009 - 02:37
User Badges:
  • Super Bronze, 10000 points or more

Unsure I understand the topology you're describing.

For gateway load balancing, rather than HSRP, GLBP might be an option.

For link load balancing, EIGRP support unequal cost load balancing. Also, if supported on you platforms, OER/PfR supports dynamic load balancing.

subra4u Mon, 04/20/2009 - 02:52
User Badges:

Thx 4 ur response.

Let me explain a bit more about the site A. The site A will need

a) Gateway Redundancy

b) utilize both the links for sending traffic i.e at any point of time there will be atleast 150+ Mbps traffic.

If we enable HSRP and IGP, it will be either asymmetric LB or restricting it to single link altering metric values. we would like to use both the links combined and also achieve gateway redundancy.

What is the best of doing it.

Note: This circuit is a P2P private circuit



Joseph W. Doherty Mon, 04/20/2009 - 03:12
User Badges:
  • Super Bronze, 10000 points or more

Gateway redundancy can be addressed by HSRP, GLBP or VRRP. Of these, GLBP also offers options to load balance gateways.

Unfortunately, still unclear (at least to me) what you'll have for a WAN topology. Depending on topology and IGP, you're not always restricted to single link LB. Even for IGPs that don't offer LB across unequal cost paths, it's sometimes possible to cost paths such that LB can be achieved.

It might help if you could provide a simple topology diagram and note against it what you're trying to accomplish.

subra4u Mon, 04/20/2009 - 04:08
User Badges:


Its a simple topology Site A has 2 X 100 Mbps P2P links 1 connecting to Site B and 1 connecting to Site C. Site B and C are connecting using a 1000 Mbps P2P link.



Giuseppe Larosa Mon, 04/20/2009 - 04:29
User Badges:
  • Super Silver, 17500 points or more
  • Hall of Fame,

    Founding Member

Hello Subra,

whatever protocol you use for first hop redundancy any IGP can give you load balancing A to B and B to A over the two parallel 100 Mbps.

you cannot load balance traffic A to B via also link to C, unless you use EIGRP with the variance command.

You should be fine with default settings with an IGP.

Hope to help


subra4u Mon, 04/20/2009 - 06:30
User Badges:

Thanks for the Input. My doubt is when I run HSRP or VRRP the active L3 device will send the traffic out all the time and the return traffic may change depending upon on the metric. How do I get the traffic out on both the links. I am not very confident with the GLBP on 2 different links (different latency) as i am planning to run voice or RTP across. Please suggest an idea.




This Discussion