AIX 5.3, DS4300 and redundancy over multiple HBA's - Aix

This is a discussion on AIX 5.3, DS4300 and redundancy over multiple HBA's - Aix ; Good day everyone, I just made another try on establishing redundant FC link for an AIX (actually - VIOS) host connected to DS4300 system through a SAN switch (2005-H16). This time it worked, but I had to connect each HBS ...

+ Reply to Thread
Results 1 to 3 of 3

Thread: AIX 5.3, DS4300 and redundancy over multiple HBA's

  1. AIX 5.3, DS4300 and redundancy over multiple HBA's

    Good day everyone,

    I just made another try on establishing redundant FC link for an AIX (actually -
    VIOS) host connected to DS4300 system through a SAN switch (2005-H16). This time
    it worked, but I had to connect each HBS on the VIO LPAR exactly to one
    controller on the DS4300. Connection diagram is like this:

    VIOS|->fcs0->H16 switch(ZONE_A: fcs0 and DS4300 controller A)->DS4300_A
    LPAR|->fcs1->H16 switch(ZONE_B: fcs1 and DS4300 controller B)->DS4300_B

    this seems to be as per 'DS4000 Best practices' redbook.

    Now the question - is that a REDUNDANT configuration? Let's say, a fiber (or
    HBA) fails connecting fcs0 to a switch and then a fiber fails connecting
    DS4300_B to a switch. In this situation there will be no path to LUNs. According
    to same 'Best Practices' redbook, every darX device cannot have more than2 dacX
    devices attached. I have checked this (by creating a single ZONE on the switch
    and including fcs1, fcs0, DS4300_A and DS4300_B ports in it) and indeed this
    configuration does not work as redundant - there is no switchover during
    failure. Windows hosts work fine in such case and got a redundant connection.

    Am I missing something or you simply cannot create a fully redundant linkto
    DS4300 SAN using 2 HBA's?


    Alex

  2. Re: AIX 5.3, DS4300 and redundancy over multiple HBA's

    On Sep 26, 5:51*am, AlexB wrote:
    > Good day everyone,
    >
    > I just made another try on establishing redundant FC link for an AIX (actually -
    > VIOS) host connected to DS4300 system through a SAN switch (2005-H16). This time
    > it worked, but I had to connect each HBS on the VIO LPAR exactly to one
    > controller on the DS4300. Connection diagram is like this:
    >
    > VIOS|->fcs0->H16 switch(ZONE_A: fcs0 and DS4300 controller A)->DS4300_A
    > LPAR|->fcs1->H16 switch(ZONE_B: fcs1 and DS4300 controller B)->DS4300_B
    >
    > this seems to be as per 'DS4000 Best practices' redbook.
    >
    > Now the question - is that a REDUNDANT configuration? Let's say, a fiber (or
    > HBA) fails connecting fcs0 to a switch and then a fiber fails connecting
    > DS4300_B to a switch. In this situation there will be no path to LUNs. According
    > to same 'Best Practices' redbook, every darX device cannot have more than2 dacX
    > devices attached. I have checked this (by creating a single ZONE on the switch
    > and including fcs1, fcs0, DS4300_A and DS4300_B ports in it) and indeed this
    > configuration does not work as redundant - there is no switchover during
    > failure. Windows hosts work fine in such case and got a redundant connection.
    >
    > Am I missing something or you simply cannot create a fully redundant linkto
    > DS4300 SAN using 2 HBA's?
    >
    > Alex

    Try to 'cascade' your switches creating a ILS between two ports in
    them

  3. Re: AIX 5.3, DS4300 and redundancy over multiple HBA's

    On Sun, 28 Sep 2008 18:06:17 -0700 (PDT), ColombianJoker
    wrote:

    >On Sep 26, 5:51šam, AlexB wrote:
    >> Good day everyone,
    >>
    >> I just made another try on establishing redundant FC link for an AIX (actually -
    >> VIOS) host connected to DS4300 system through a SAN switch (2005-H16).This time
    >> it worked, but I had to connect each HBS on the VIO LPAR exactly to one
    >> controller on the DS4300. Connection diagram is like this:
    >>
    >> VIOS|->fcs0->H16 switch(ZONE_A: fcs0 and DS4300 controller A)->DS4300_A
    >> LPAR|->fcs1->H16 switch(ZONE_B: fcs1 and DS4300 controller B)->DS4300_B
    >>
    >> this seems to be as per 'DS4000 Best practices' redbook.
    >>
    >> Now the question - is that a REDUNDANT configuration? Let's say, a fiber (or
    >> HBA) fails connecting fcs0 to a switch and then a fiber fails connecting
    >> DS4300_B to a switch. In this situation there will be no path to LUNs.According
    >> to same 'Best Practices' redbook, every darX device cannot have more than 2 dacX
    >> devices attached. I have checked this (by creating a single ZONE on the switch
    >> and including fcs1, fcs0, DS4300_A and DS4300_B ports in it) and indeed this
    >> configuration does not work as redundant - there is no switchover during
    >> failure. Windows hosts work fine in such case and got a redundant connection.
    >>
    >> Am I missing something or you simply cannot create a fully redundant link to
    >> DS4300 SAN using 2 HBA's?
    >>
    >> Alex

    >Try to 'cascade' your switches creating a ILS between two ports in
    >them


    I belive it is not going to work, as I am obliged to put HBA and one portat
    DS4300 in one zone (otherwise - more than 2 dacX devices appear on a darX).
    Cascading switches will yield same result (although will add protection from
    switch failure). Seems that I will need to use an extra host license (to create
    another darX device) and then use either 4 HBAs on the host or create zone with
    and fxsX port and ports A/B from DS4300.

    Heard somewhere that latest DS4K firmware (from v6.19) supports 'mpio' way of
    doing things (as opposed to 'RDAC' way), so I will try it.

    Windows drivers seem to use both 'rdac' and 'mpio' at the same time somehow.

    Alex

+ Reply to Thread