SD-WAN

  • 1.  Redundancy Configuration

    Posted 03-09-2020 06:59
    Hi Team,

    Is it mandatory to configure redundancy-group configuration when both nodes are acting as Active/standby?


    ------------------------------
    Bhaskar Gaddam
    Solution Architect
    (91973) 024-1175
    ------------------------------


  • 2.  RE: Redundancy Configuration

     
    Posted 03-10-2020 09:15
    Hi Bhaskar, the redundancy-group configuration is not mandatory except for deployments where two 128T nodes are operating in an active/standby capacity and there is no fabric link between them.

    The redundancy-group configuration lets an administrator define groups of interfaces that "share fate" -- that is, when one of them fails, it will treat all of them as down for the purposes of interface leadership. (A secondary attribute lets you define the cost of the group, so you can configure administrative preference for which group should be active in the "sunny day" scenario.)

    Assume a simple deployment with two nodes (node1 and node2), a single LAN shared between the two nodes, and a single WAN shared between the two nodes. You've configured no redundancy groups and no fabric interface. Both interfaces are active on node1. If the LAN interface fails, node2 will take over the LAN interface. However, the active WAN interface is still on node1. Packets will arrive on node2 and have no way to reach the active WAN interface on node1.

    To address this type of failure, you have two choices.
    1. You can configure a redundancy-group. This will "group" the LAN and WAN together, such that if the LAN fails on node1, it will consider the WAN interface failed as well -- and switch both of them over to node2.
    2. You can configure a fabric interface. Thus, if LAN fails on node1 the packets will arrive on node2, traverse the fabric interface, and egress the WAN interface on node1. (Additionally, the fabric interface protects you against the dual failure where LAN fails on node1 and WAN fails on node2.)

    You can also configure BOTH of these at the same time. This is done primarily to keep ingress and egress traffic on the same node in the event of a LAN or WAN failure (and not use the fabric interface). So a LAN interface failure on node1 causes node2 to assert itself for both LAN and WAN. But in the event of a WAN interface failure on node2, you'll have a valid path in node2 and out node1.

    I hope this clears some doubts up.

    ------------------------------
    pt.
    ------------------------------



  • 3.  RE: Redundancy Configuration

    Posted 03-11-2020 13:32
    Thanks PT.
    This really helps.

    ------------------------------
    Bhaskar Gaddam
    Solution Architect
    (91973) 024-1175
    ------------------------------