Wired Intelligent Edge

 View Only
  • 1.  ArubaCX 8325 question - lag vs mclag

    Posted Feb 28, 2022 12:57 PM
    Edited by RedX Feb 28, 2022 01:16 PM
    I have two 8325s setup with VSX between the two - all is happy and working.  There are no L3 interfaces on this VSX stack. My question is, I want to connect a QNAP NAS with LACP and I assume I will just want to setup a traditional lag interface versus an mclag, correct?  I know I would use an mclag to another switch, but when it comes to a network device like a NAS or server, would an mclag work?


  • 2.  RE: ArubaCX 8325 question - lag vs mclag
    Best Answer

    Posted Feb 28, 2022 01:17 PM
    Hi Wes, on the contrary...if your NAS is going to be connected to VSX Primary and Secondary (say one physical link from NAS to VSX Primary and another physical link from NAS to VSX Secondary) you should use VSX LAGs with LACP so Multi-Chassis LAG with LACP. Traditional LAGs (Non Protocol or LACP) are going to be used when the connections terminate against one VSX member only (and not against both).

    We have a QNAP TS-1283XU-RP exactly connected that way (Fiber Optic links, 10G+10G).

    ------------------------------
    Davide Poletto
    ------------------------------



  • 3.  RE: ArubaCX 8325 question - lag vs mclag

    Posted Feb 28, 2022 01:22 PM
    Edited by RedX Feb 28, 2022 01:23 PM
    Ah ok, thank you for the clarification!  So very few scenarios I can think (minus the VSX configuration) where you would configure a lag (or non-mclag) interface in a VSX environment.  

    ------------------------------
    Wes Wilson
    ------------------------------



  • 4.  RE: ArubaCX 8325 question - lag vs mclag

    Posted Feb 28, 2022 02:23 PM
    Yes, exactly. Basically if a VSX configuration is used it is also supposed that that downstream (and also upstream) peers - so Switches, Servers and so on - are going to be connected to both VSX members concurrently by means of a LAG on their sides (preferentially with LACP as the links aggregation control protocol), this scenario implies that VSX has more VSX LAGs than non-VSX LAGs on its side...and the latter cases (non-VSX LAGs) are used for uncommon cases (OTOH if you have a peer which is LAG LACP capable why not connecting it to both VSX members?).

    ------------------------------
    Davide Poletto
    ------------------------------