Wired Intelligent Edge

 View Only
last person joined: 23 hours ago 

Bring performance and reliability to your network with the HPE Aruba Networking Core, Aggregation, and Access layer switches. Discuss the latest features and functionality of your switching devices, and find ways to improve security across your network to bring together a mobile-first solution
Expand all | Collapse all

How can i connect 4x8325? VSF and VSX

This thread has been viewed 55 times
  • 1.  How can i connect 4x8325? VSF and VSX

    Posted Jan 31, 2023 10:48 AM
    I have to build a core with 4 8325s, 2 in one location and the other two in another location. Is it possible to make a Stack between 2 using VSF? Thus we would only have 2 logical units with which VSX could be mounted.
    In the figure below you can see what I mean.

  • 2.  RE: How can i connect 4x8325? VSF and VSX

    Posted Jan 31, 2023 11:16 AM
    The 8325 does not support VSF.  Your correct configuration here would be two VSX clusters.

  • 3.  RE: How can i connect 4x8325? VSF and VSX

    Posted Jan 31, 2023 11:24 AM
    But then i have to create three vsx stack? 
    I am reading that VSX only can on two devices

  • 4.  RE: How can i connect 4x8325? VSF and VSX

    Posted Jan 31, 2023 11:47 AM
    Hi, why three VSX Stacks? you have to create two VSX (on each site the VSX shall be made by 2 Aruba 8325) and interconnect them "back-to-back" (using Multi-Chassis LAGs as known as VSX LAGs).

    Have a look at VSX Best Practices Guide, at the end there is an example where two VSX, each one operating in a specific site, are interconnected via four physical links through VSX LAGs.

  • 5.  RE: How can i connect 4x8325? VSF and VSX

    Posted Feb 01, 2023 05:19 AM
    So if I want to use a LAG between them towards other wiring centers, would it be possible?

  • 6.  RE: How can i connect 4x8325? VSF and VSX

    Posted Feb 01, 2023 06:54 AM
    Not in the way you draw those downlinks (uplinks).

    The point is: interlinking two VSX(s) together doesn't create a VSX of VSX(s) (a cluster of clusters), you just have two VSX(s) interconnected together (on each VSX you will have to setup MC-LAGs, one on each VSX member of that particular VSX cluster hosted on that particular site, these two MC-LAGs of the VSX on the left site will form the LAG that will be used to connect to the corresponding LAG formed on the VSX on the right side (again, in that side the VSX will have its MC-LAGs, one per VSX member, to connect to the left side).

    Given the above you have just interconnected the left VSX with the right VSX. If you have other devices those ones need to connect (and terminate their links) to each VSX and only there...you can't go from, example, CC2 and VSX on the left with one downlink and VSX on the right with the other downlink (as pictured)...but you can go from CC2 to VSX on the left (or on the right) with both of its downlinks spreading into VSX members of the same VSX (say one CC2 downlink to left VSX top member and the other downlink to the left VSX bottom member).

    Technically - having a lot of links between the sites - you can also create other downlinks to the right VSX...but...doing so you will form a loop (CC2 <-> Left VSX <-> Right VSX <-> CC2) and I don't believe that is something you want architecturally speaking.

    So, to recap, if from CC2 you have a LAG, its downlinks to VSX clusters (on Left and Right sites) can go from:

    - CC2 to, respectively, concurrently Left VSX Top and Bottom members <- OK
    - CC2 to, respectively, any Left VSX member AND any Right VSX member <- Not OK

    The same is valid for CC3 and CC4, always speaking about a LAG (and its member links) going to your two sites.

    A side note: you can connect CC2 via standalone downlinks (non LAGged) to - say - one member of Left VSX and one member of Right VSX but expect that one of the two downlinks will be blocked (since that's a loop)...so, in the end, you will have a CC2 connected and forwarding to only one site at time (forwarding will not happen concurrently)...this if STP is working correctly (remember that I assumed that, for this particular case, you are not dealing with a LAG from CC2...).

  • 7.  RE: How can i connect 4x8325? VSF and VSX

    Posted Feb 01, 2023 07:09 AM

    Hi parnassus,

    So I have the following problem: with only two computers I don't have enough links for all my cabling centers. For this very reason I am going to use 4 teams. Is there no way for these four computers to have simultaneous redundancy? Can you only make a stack of two teams? If you could provide me with the best practices VSX example, I would appreciate it.


  • 8.  RE: How can i connect 4x8325? VSF and VSX
    Best Answer

    Posted Feb 01, 2023 07:55 AM
    See here, page 106 "APPENDIX F – VLAN extension between two VSX clusters".

    Let me suppose you have an Host hosted on a site (let's say Site-L on the left side) and you want it to be concurrently connected to both sites (Left and Right) then, supposing you have plenty of physical links (at least two internally at Site-L and at least two going from Site-L to Site-R) then you must connect that Host both to Site-L and Site-R BUT, as written, this connection can't be done with just a single LAG (from the Host standpoint) because Host's LAG member links need to terminate into a common logical entity (against a single VSX cluster, not against both VSX clusters).

    At best you can imagine that that Host has two separate LAGs, one going to Site-L VSX cluster and the other going to Site-R VSX cluster...but one of these two LAGs will be blocked by STP (Host - LAG1 - VSX at Site-L - VSX at Site-R - LAG2 - Host <- it's a loop) and so it seems quite questionable as approach.

    Another case - NOT involving LAG - would be the one where you have an Host with four standalone links (and each link is connected to each VSX member of any site)...but I don't know why you would desire such type of connectivity (Virtualization Platform?).

    Edit: A VSX deployment uses ONLY TWO member switch. VSF and VSX are different technologies, you can't mix them (a VSF of VSX or a VSF of VSF...VSF is VSF - from minimum two members to "n" members depending on various factors (OS, as example) - and VSX is VSX = 2 members).

  • 9.  RE: How can i connect 4x8325? VSF and VSX

    Posted Feb 03, 2023 03:27 AM
    Thanks for your reply parnassus,

    For us, the unique solution will be make two vsx in the two sites and then we will   create a MLAG from a team from one site and another team from the other site. It would only be two cables instead of throwing two for each VSX (total 4). It is the only way we have to make this connection and have enough ports for all the wiring centers.

    The truth is that it is quite annoying that these 8325 teams do not have a way of stacking for more than 2 devices.


  • 10.  RE: How can i connect 4x8325? VSF and VSX

    Posted Feb 03, 2023 09:44 AM
    Hi, the reality is that VSX is not a "stacking" technology but a "clustering" technology where there is a Primary VSX Member Switch synchronizing to a Secondary VSX Member Switch in a Active (Pri) to Active (Sec) direction (see here): VSX was engineered to require a 1 to 1 relationship, it's not based on a relationship (and it can't be) between 1 to many (typically the switch with a Conductor role is the 1 controlling the "many" and these ones have roles like Standby and Members). The 1 to many applies when you deal with Hardware (Backplane) or Software (Frontplane) stacking solutions, like when you deal with a Frontplane stacking solution deployed through VSF using ArubaOS-Switch based switch series like the Aruba 2930F, Aruba 2930M, 5400R zl2 or ArubaOS-CX based switch series like the Aruba CX6200 and Aruba CX 6300 or, alternatively, when you deal with a Backplane stacking solution - so with specific Stacking Hardware involved - deployed on ArubaOS-Switch based switch series like the Aruba 2920 or Aruba 3810M).