Switching

 View Only
last person joined: yesterday 

Ask questions and share experiences about EX and QFX portfolios and all switching solutions across your data center, campus, and branch locations.

  • 1.  Has anyone actually stacked 10 EX4400's successfully?

    Posted 02-27-2023 15:35

    Hey all, we are working on a switch from Cisco to Juniper.  We are looking to see if anyone has actually stacked 10) EX 4400's together successfully.  I know documentation says it will work to stack that many, but I am wondering if anyone has stacked the max amount and if there were any gotchas or performance issues.



    ------------------------------
    Randy Shulse
    ------------------------------


  • 2.  RE: Has anyone actually stacked 10 EX4400's successfully?

    Posted 03-03-2023 16:42

    Hi Randy,

    over the years I deployed tons of VCs and from personal experience I can only strongly advise you to never go beyond a 6-Member VC. A 10 Member VC is doable (and I had multiple running) but besides a commit taking forever there are some „challenges" during upgrades or partial failures. If your VC has an issue then ALL Members can be affected and the more Members you have the bigger the impact will be. If you plan to introduce your 7th Member it's better to split the Members into 2 VCs.

    Hope this helps



    ------------------------------
    Christian Scholz
    Juniper Networks Ambassador | JNCIE-SEC #374
    Mail: chs@ip4.de
    Blog: jncie.eu | Twitter: @chsjuniper | YT-Channel: netchron
    ------------------------------

    Original Message


  • 3.  RE: Has anyone actually stacked 10 EX4400's successfully?

    Posted 03-04-2023 02:58
    I have the same experience (not with Juniper but another brand). Keep in
    mind that if you're running stuff like LLDP, LACP and STP the control
    plane needs to send frequent payloads on a lot of ports, in some cases
    almost all of them. The CPU of a switch is not designed for that and
    you'll experience lots of delays and, in my case, even lots of SNMP
    fetching timeouts.

    I'd recommend to do either one of:

    - break up in smallers stacks, not going beyond four members actually.
    This can be done if you have some spare uplink media available (fibers
    etc.) and also enough ports available on the distribution or core layer.
    In my situation that was always true.
    - start using a chassis, they're designed for handling large quantities
    of access ports.


    Original Message