Issues with Server ALB (Adaptive Load Balancing)

  • 0
  • 1
  • Problem
  • Updated 1 year ago
  • Not a Problem
Hi There,

I'm trying to get more information about ALB. We've got some servers using it and we're seeing odd behaviour.

The scenario is, two edge switches in an eaps ring with two cores, switches on the edge are 450-G2's, code on the edge is and the cores are BD8810s and the code is the same.

There is an EAPS ring with the master node at the edge, where the master node is blocking a leg to one of the cores, therefore meaning the two edge switches can talk between eachother without eaps interfering.

The issue is, the hosts on the edge switches can't talk correctly to the rest of the network when the ring is complete, i have to physically disable a leg to one of the cores to get it to work correctly. it doesn't matter which leg to either core.

I have read that ALB doesnt require any special switch configuration and that it should support two nics from the same server, one in each of the switches to allow load balancing and switch resilience.

hope that makes sense.

anyone had any experience with this? would it be better to go the LACP route?

we would use MLAG to pair the switches in essence and configure LACP on the servers.

Photo of Ian Broadway

Ian Broadway

  • 1,742 Points 1k badge 2x thumb

Posted 1 year ago

  • 0
  • 1
Photo of Erik Auerswald

Erik Auerswald, Embassador

  • 13,772 Points 10k badge 2x thumb
Hi Ian,

what is ALB and how does it work? Does it use multicast, which might interact with IGMP snooping on the switches? Do you have any links to documentation?