Brocade Fibre Channel Networking Community

Expand all | Collapse all

Virtual Fabric

  • 1.  Virtual Fabric

    Posted 10-31-2019 12:08 PM
    Hi, we are using the Brocade X-6 models and we configured VF on each switch.
    ISL connectivity between edge to edge switches but am unable to check the ISL ports health .
    I tried to run the portshow -i index number and portshow slot/port.

    Portshow -i getting error ---->index number do not work in logical switch.

    in switch show ISL port number is showing below
    Index slot Port Address Media.......
    448    -1    448  ----------------------------------E-port

    May i know how to check?

    ------------------------------
    Sankar Gulla
    ------------------------------


  • 2.  RE: Virtual Fabric

    Posted 28 days ago
    A negative index indicates that it is a virtual port. For example, if you are using a base switch for ISLs, there is no real E-Port in your logical switch so the virtual E-Port appears with -1 as the index.

    ------------------------------
    Field Applications Engineer
    Broadcom
    ------------------------------



  • 3.  RE: Virtual Fabric

    Posted 24 days ago
    Thanks Jack for reply. How can i trace that this logical switch is going through which port from base switch?


  • 4.  RE: Virtual Fabric

    Posted 23 days ago
    To see where the ISLs are, just setcontext n, where n if the FID number of the base switch. islshow, switchshow, trunkshow, ...

    The trunking and routing algorithms in the base switch are the same as they are in a logical switch with direct ISLs. Since most environments have redundant paths.

    I don't believe traffic shaping (traffic isolation zones) is supported when using a base switch so if this is something you need, let me know and I'll research it a little further.

    This means the path for each frame is determined by:

    • Fastest Shortest Path First (FSPF).
      • Fibre channel assigns a cost to available paths based on the number of hops and link speed
      • A lower cost means fewer hops and higher speed.
    • Exchange based routing (EBR)
      • If there are multiple paths at the same lowest cost, the EBR algorithm will choose the least used path. The EBR algorithm is a little more complex but this description is adequate for purposes of this discussion.
      • A fibre channel exchange will always complete on the same path.
      • A trunk is considered a single path.
    • Brocade trunking algorithm
      • This algorithm sends frames on certain physical links and re-assembles them in order at the remote end effectively making all ISLs in a single trunk appear as a single large ISL


    ------------------------------
    Field Applications Engineer
    Broadcom
    ------------------------------



  • 5.  RE: Virtual Fabric

    Posted 20 days ago
    Hi Jack, Thank you.

    We've multiple physical chassis and configured Logical ISL between same Fab IDs via base switch.
    do we've any possibility to communicate through this ISLs if FIDs are not same? if yes, may i know the process?

    We've isolated the traffic by segregating the ports into different logical switches. do we've any other way to segregate the ISL traffic?

    Thank you.


  • 6.  RE: Virtual Fabric

    Posted 20 days ago
    ------>>>>do we've any possibility to communicate through this ISLs if FIDs are not same? if yes, may i know the process?

    via FCR

    ------------------------------
    TechHelp24
    ------------------------------



  • 7.  RE: Virtual Fabric

    Posted 19 days ago
    Although FID checking can be disabled, I don't believe it will work in your case. I do not have intimate knowledge of how the base switches work so my response is based on my understanding of how it works.

    Base Switch

    The base switches must have the same FID which cannot be the same as the FIDs for any of the logical fabrics using the base switch. The base switch effectively acts a fibre channel router in that it establishes routes between ISL connections from like FIDs through the interconnected base switches.

    For example, assume there are two chassis with a base switch, FID 127, and two logical switches, FIDs 1 & 2, configured to allow XISL (XISL is the term used for ISLs that traverse the base switch). FID 1 in chassis A has domain ID (DID) 10 and FID 1 in chassis B has DID 11. Similarly, FID 2 has DIDs 20 & 21. The base switch will establish paths between the two logical switches with FID 1 because they have the same FID and paths between the two logical switches with FID 2 because they have the same FID. The result is two isolated logical fabrics, FID 1 (DIDs 10 & 11), and FID 2 (DIDs 20 &21) both sharing the same physical ISLs between the base switches.

    Disabling FID Check

    The FID check can be disabled with the CLI command configure, then answer "Yes" to "Fabric Parameters". I don't recall the exact verbiage but one of the prompts will ask if you want to disabled FID checking. The default is that FID checking is enabled.

    I have never recommended disabling FID checking in a production environment and cannot think of a circumstance where I would. Disabling FID checking makes sense in a lab environment because you can create multiple logical switches and ISL them together to form test fabrics while limiting lab costs to a single physical switch.

    Disabling FID Checking With the Base Switch

    I have no idea what the base switch would do if FID check was disabled. I have never seen this in any stated supported topology and have no knowledge of anyone ever attempting it. For this reason, if it's important to you, you will need to raise this question through your support organization.

    My assumption is that FOS would check for this setting and and reject the connection. It's theoretically possible to design and implement an algorithm that would treat all FID connections using the base switch with FID checking disabled as one large FID but having been both an engineering manager and product manager in past careers, I can't imagine anyone justifying the extra development and test cost to do so. Keep in mind that all supported configurations must go through regression testing for each version of FOS so I'm assuming test is the the more prohibitive cost.

    A Possible Work-Around

    You can have the logical fabrics configured as in the previous example using the base switch for transport while having another logical fabric with FID checking and Allow XISL use disabled with ISLs connected between them.

    It's not clear to me what the ultimate goal is. If you can be more explicit on what you are trying to accomplish, perhaps I can provide a better alternate solution.

    ------------------------------
    Field Applications Engineer
    Broadcom
    ------------------------------