SAN Compatibility Question - Veritas Cluster Server

This is a discussion on SAN Compatibility Question - Veritas Cluster Server ; Will Veritas Cluster Server work in this situation? I have an E4500 and a Sun-Fire 280R. Both are attached to the SAN (IBM FastT-700's and Brocade Silkworm switches) and both can see the same disks. (I can deport/import vxvm volume ...

+ Reply to Thread
Results 1 to 4 of 4

Thread: SAN Compatibility Question

  1. SAN Compatibility Question


    Will Veritas Cluster Server work in this situation?

    I have an E4500 and a Sun-Fire 280R. Both are attached to the SAN (IBM
    FastT-700's and Brocade Silkworm switches) and both can see the same
    disks. (I can deport/import vxvm volume groups between the two systems)
    Based on this, can I assume that cluster server will work?

    This is going to be a failover Oracle database based on the following
    hardware:

    E4500:
    1 quad NIC (for heartbeets)
    2 gigabit NICs
    2 JNI HBA's (each attached to different Brocade switches, not redundant
    pathing)

    280R;
    1 quad NIC
    2 gigabit NICs
    2 JNI HBA's

    Both will be running DBE-HA for Oracle 3.5MP1.

    My only concern is this IBM RDAC resolution layer that is required. We
    are using IBM Fast-T 700 Storage Controllers with Brocade Silkworm
    Fabric Switches. Initially each disk must be visible under two different
    SCSI id's because the way the FastT-700's are designed, once the RDAC
    layer is installed, each disk is represented by a single SCSI ID and the
    RDAC layer handles this. (since a disk owned by controller A in FastT-1
    will appear as c0t1d0 and a disk owned by controller B in FastT-1 will
    appear as c0t2d0, because each of the controllers in each of the
    fastT-700's must be persistently binded to a specific SCSI ID)


    Thanks for any help,
    CC


    --

    Chuck Carson
    Sr. Systems Engineer
    Syrrx, Inc.
    10410 Science Center Drive
    San Diego, CA 92121


  2. Re: SAN Compatibility Question


    Hello,

    with DMP there is no problem !!!
    if type "vxdisk list c0t1d0" you will see at the end that there are 4 paths
    to access to this disk (c1t1d0, c1t2d0 on the JNI1 and c2t1d0, c2t2d0 on
    the JNI2)

  3. Re: SAN Compatibility Question

    Hi,

    I doubt there would be any problems. I have a customer running a
    FastT-500 and Sun maschines running VCS 2.0 in a SAN environment. For
    the FastT-500 we had to use the RDAC drivers to handle the multipathing
    feature. In our case the RDAC driver "hides" the physical devices and
    reports a Nexus device back to the OS, hence if you run format you will
    only see one device. I would guess that the FastT-700 works in the same way.
    And if you manually can import and deport the diskgroups between the
    nodes, VCS will be able to handle it.

    /Peter

    Rhugga wrote:
    >
    > Will Veritas Cluster Server work in this situation?
    >
    > I have an E4500 and a Sun-Fire 280R. Both are attached to the SAN (IBM
    > FastT-700's and Brocade Silkworm switches) and both can see the same
    > disks. (I can deport/import vxvm volume groups between the two systems)
    > Based on this, can I assume that cluster server will work?
    >
    > This is going to be a failover Oracle database based on the following
    > hardware:
    >
    > E4500:
    > 1 quad NIC (for heartbeets)
    > 2 gigabit NICs
    > 2 JNI HBA's (each attached to different Brocade switches, not redundant
    > pathing)
    >
    > 280R;
    > 1 quad NIC
    > 2 gigabit NICs
    > 2 JNI HBA's
    >
    > Both will be running DBE-HA for Oracle 3.5MP1.
    >
    > My only concern is this IBM RDAC resolution layer that is required. We
    > are using IBM Fast-T 700 Storage Controllers with Brocade Silkworm
    > Fabric Switches. Initially each disk must be visible under two different
    > SCSI id's because the way the FastT-700's are designed, once the RDAC
    > layer is installed, each disk is represented by a single SCSI ID and the
    > RDAC layer handles this. (since a disk owned by controller A in FastT-1
    > will appear as c0t1d0 and a disk owned by controller B in FastT-1 will
    > appear as c0t2d0, because each of the controllers in each of the
    > fastT-700's must be persistently binded to a specific SCSI ID)
    >
    >
    > Thanks for any help,
    > CC
    >
    >



  4. Re: SAN Compatibility Question

    BTW,

    We couldn't use vxdmp with the FastT-500. It just doesn't work. vxdmp
    needs to see the same serialnumber from the different I/O paths, and the
    FastT-500 doesn't give you the same serialnumber on the different I/O
    paths. That's why we had to use the RDAC drivers. I don't know if this
    is the case with the FastT-700, but I guess it is.

    /P

    Peter Sevborn wrote:
    > Hi,
    >
    > I doubt there would be any problems. I have a customer running a
    > FastT-500 and Sun maschines running VCS 2.0 in a SAN environment. For
    > the FastT-500 we had to use the RDAC drivers to handle the multipathing
    > feature. In our case the RDAC driver "hides" the physical devices and
    > reports a Nexus device back to the OS, hence if you run format you will
    > only see one device. I would guess that the FastT-700 works in the same
    > way.
    > And if you manually can import and deport the diskgroups between the
    > nodes, VCS will be able to handle it.
    >
    > /Peter
    >
    > Rhugga wrote:
    >
    >>
    >> Will Veritas Cluster Server work in this situation?
    >>
    >> I have an E4500 and a Sun-Fire 280R. Both are attached to the SAN (IBM
    >> FastT-700's and Brocade Silkworm switches) and both can see the same
    >> disks. (I can deport/import vxvm volume groups between the two
    >> systems) Based on this, can I assume that cluster server will work?
    >>
    >> This is going to be a failover Oracle database based on the following
    >> hardware:
    >>
    >> E4500:
    >> 1 quad NIC (for heartbeets)
    >> 2 gigabit NICs
    >> 2 JNI HBA's (each attached to different Brocade switches, not
    >> redundant pathing)
    >>
    >> 280R;
    >> 1 quad NIC
    >> 2 gigabit NICs
    >> 2 JNI HBA's
    >>
    >> Both will be running DBE-HA for Oracle 3.5MP1.
    >>
    >> My only concern is this IBM RDAC resolution layer that is required. We
    >> are using IBM Fast-T 700 Storage Controllers with Brocade Silkworm
    >> Fabric Switches. Initially each disk must be visible under two
    >> different SCSI id's because the way the FastT-700's are designed, once
    >> the RDAC layer is installed, each disk is represented by a single SCSI
    >> ID and the RDAC layer handles this. (since a disk owned by controller
    >> A in FastT-1 will appear as c0t1d0 and a disk owned by controller B in
    >> FastT-1 will appear as c0t2d0, because each of the controllers in each
    >> of the fastT-700's must be persistently binded to a specific SCSI ID)
    >>
    >>
    >> Thanks for any help,
    >> CC
    >>
    >>

    >



+ Reply to Thread