Routing

 View Only
last person joined: 4 days ago 

Ask questions and share experiences about ACX Series, CTP Series, MX Series, PTX Series, SSR Series, JRR Series, and all things routing, including portfolios and protocols.

  • 1.  Question on VPLS

     
    Posted 11-21-2023 15:50

    Hi.

    Does an MX PE transparently forward customer layer 2 control frames such as LACP, BPDUs, OAM, etc across VPLS pseudowires by default?

    Thanks,

    Deepak



  • 2.  RE: Question on VPLS
    Best Answer

    Posted 11-22-2023 10:51
    Edited by djadhav 12-13-2023 00:49

    I'll share what I'm seeing with STP BPDU's over VPLS, but I can't speak to LACP or OAM.

    I started up a manual ldp-based vpls (not bgp ad) in my eve-ng lab using vmx and vqfx.  I pulled some packets using wireshark to see what some of this looks like.  I see STP being sent from root bridge (vqfx-9) to the other bridge (vqfx-5) via the vpls provided by the vMX PE's.  Something that looks very interesting and strange is that the STP BPDU traffic sent from root to non-root bridge inside the vpls pseudowire looks like this. 

    In the core of the MPLS network between the PE's, I did a wireshark capture at a central P-node (LSR) in order to see all the possible layers of encapsulation.  Of course there I see the lowest level natural ethernet header between PE-P nodes, and the 2-label stack for core label and l2vpn-vpls label, then the PW control header and then a IEEE 802.3 header which I'm thinking is created by the VPLS process in the PE's.  Deeper than that seems to be the actual STP BPDU as created from the v-qfx CE's.  In that IEEE 802.3 Ethernet header encapsulated inside the PW header, I see the following source and destination mac addresses, but strangely re-arranged like this. 

    For reference, I'll tell you that the natural (real) source mac of the root bridge vqfx-9 is ...
    02:05:86:71:2a:02

    but inside the pseudowire, i see the ieee 802.3 destination and source mac like this...

    cc:cd:02:05:86:71 - destination mac
    2a:02:81:00:e0:11 - source mac

    What it appears to be is, the original vqfx-9 root bridge mac is broken apart, where the first 4 bytes (02:05:86:71) is prepended with cc:cd to form the destination mac used inside of the pseudowire as a new ethernet header, and then the last two bytes of the root bridge natural mac address (2a:02) is prepended to what is seen as 81:00:e0:11 to form the new source mac inside the pseudowire as 2a:02:81:00:e0:11.  I'll guess 81:00 is representing 0x8100 as standard ethertype for single tagged 802.1Q frames.  e0?  I don't know what e0 could be.  The trailing 11 looks like it might be the hex equivalent for the carried vlan number, decimal 17

    That's so strange!  I'd love to see a document or web link explaining this.

    in this weblink you will see references to "cc:cd" but usually at the end of the mac address, not at the beginning as I'm observing in my lab
    https://www.juniper.net/documentation/us/en/software/junos/multicast-l2/topics/topic-map/layer-2-protocol-tunneling.html



    ------------------------------
    - Aaron
    ------------------------------

    Original Message


  • 3.  RE: Question on VPLS

     
    Posted 12-13-2023 00:51

    Thanks Aaron. for the detailed reply.

    Cheers!


    Original Message