Comware

 View Only
  • 1.  Rapid spanning tree topology

    Posted Jun 25, 2003 03:30 AM
    Good morning,

    I would like to put in place a topology with Rapid spanning tree like this:

    Room1 Room2
    H4108GL1----------------------HP4108GL2
    HP2524_1-----------------------HP2524_2

    HP4108gl1 and HP2524_1 are connected using a trunk connection. The Same for HP4108gl2 and HP2524_2.
    the two links "-----" are FO LX 1Gb/s

    The primary link would be the HP4108GL1 and HP4108GL2 link and the backup the HP2524_1 and 2 link.

    I am a quite confused with the priority I have to assign on each switch and on each port because in this case I do not know who is the root and the slave because it is a ring.

    Thank you for help

    Best regards

    Guillaume


  • 2.  RE: Rapid spanning tree topology

    Posted Jun 25, 2003 05:10 AM
    Hi, Guillaume !

    The Spanning Tree Protocol implies the switch with the lowest Bridge ID to become the root.
    The bridge ID is made up of a 2-byte priority value (a priority of a switch not a port) followed by a 6-byte MAC address.

    BR,
    Dmitry


  • 3.  RE: Rapid spanning tree topology

    Posted Jun 25, 2003 06:54 AM
    Bonjour Guillaume!

    1. 2524s are capable of RSTP? I only have 4108GLs and I know they are. During the RTSP election, the "lowest MAC" SWT becomes the root. Generally, it is not a good ideea to mess-up with RSTP parameters - so leave them at their default values.

    2. If you REALLY want a specific SWT to be the root, you modify its priority to a lower one. It will become the root, but when it will fail a new election will take place. During the election, your useful network traffic will virtually cease. If you want no elections, you "diktate" the priority for all SWTs: each with a different value.

    3. If you see traffic on both fibers, it is because RSTP is "blocking" a trunk between a 4108GL a 2524 - not necesarily a fiber (as you might wish).

    4. If configured OK and running on all SWTs, RSTP can help you a LOT. 5 days ago I needed to disconnect a fiber for maintenance, and RSTP did its job very well: I did not receive any phone call saying "My network is down!!!". So RSTP it's great not only for accidental loss-of-link, but for planned maintenance also.

    5. See "Spanning Tree Options: RSTP (802.1w)
    and STP (802.1d)" in "4100_Mgmt&Config (ed4).pdf" that comes on the CD with the 4108GLs. Pay special attention to "Optimizing the RSTP Configuration" on page 13-9.

    Cheers,
    Dan


  • 4.  RE: Rapid spanning tree topology

    Posted Jun 27, 2003 04:35 AM
    Thank you!!

    Dan, 2524s are capable of RSTP.
    I have found the PDF documentation in the CD like you said so now, I have all the informations I need to put in place my infrastructure. I hope it will run as well as you Dan...

    Thank you again. I will let you know the result when the infrastructure will be installed

    Best regards

    Guillaume



  • 5.  RE: Rapid spanning tree topology

    Posted Jun 27, 2003 10:03 AM
    Hi Guillaume,

    Why not assigning points if those tips were useful to you ?

    Tks for them :]]

    J


  • 6.  RE: Rapid spanning tree topology

    Posted Jul 01, 2003 10:51 AM
    Another Question:

    I have studied the different crash possibilities and I have a problem with one of them:

    The priority are defined like in the attached schema.

    In this case if one of the switch crash, the backup link will work.
    But If my fiber between the two HP4108GL crash, Will the backup link between the two HP2524 work?
    Because the packet on the 4108 will go on HP2524 but the fiber will be still blocked no?

    Thanks for help and sorry for my english :-]

    Guillaume


  • 7.  RE: Rapid spanning tree topology

    Posted Jul 01, 2003 10:52 AM
      |   view attached
    oups!! the attached file

    Attachment(s)

    txt
    3265.txt   703 B 1 version


  • 8.  RE: Rapid spanning tree topology

    Posted Jul 01, 2003 12:38 PM
    Hello Guillaume,

    In a RSTP architecture, if any link in a loop crash, there will be a topology change, and a new path is calculated arround this loop. So you have a blocked port in your loop. When the physical loop is brocken, the blocked port will become forwarding.

    In your picture, there is now way to guess which link will be blocked, even if you configure port priority (it is not the best paramter to use here)

    - The root Bridge should be one of your 4108 (More CPU, and I suppose it have servers on it) (spanning-tree priority 0). You can set the second 4108 has the backup root bridge (spanning-tree priority 10)
    Note that the Root bridge election is based on the bridge ID (Bridge Priority + MAC Address) contained in all BPDU.

    - With a root bridge and a backup root bridge (both 4108), the link between your 2524 should become the backup link. (Don't forget only one port of a link is blocked)

    Arnaud