IBM FlashSystem

 View Only
  • 1.  IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted Tue February 11, 2025 10:46 AM
    Edited by Sudhir BISHT Tue February 11, 2025 12:28 PM

    We've a A9000R still in production and under IBM support as well entire 2025. In few months we are going to add a new IBM FS7300. Now, to migrate data from A9000R to FS7300, mostly we will do LVM(Host Based Online Linux), ASM Balancing (Database) and vMotion (VMware). But we also have some Windows stuff. What is the easiest and best suited Array-Based Migration for this scenario. We read we can do External Storage Virtualization, as we do have Spectrum Virtualize Volume Migration in the quote. We'll not keep External Storage after the migration, so we don't need license for that like it is with any other major vendor. Is that correct. For another major vendor Storage we have here, after External Storage Virtualization, we've have to get a 'Volume Migrator' license(temp or purchased), as once the External Storage Volumes become local, we then actually migrate them to new volumes. How that part is done in FS7300. And, does here also we need to have a temp license till we complete the 'Volumes Migrations'. Please advise.

    Thanks.



  • 2.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)
    Best Answer

    Posted Tue February 11, 2025 11:20 AM
    Edited by Sudhir BISHT Thu February 13, 2025 09:51 AM

    Hi.  We just went through something similar, migrating our SVC based storage (with v7000s underneath) to new FS9500 arrays.  AIX partitions, Oracle Databases, VMware datastores and physical Linux were very simple to move as you seem to already know.  We had some "concerns" with the few physical Windows and iSeries partitions especially.  We ended up doing "reverse images" to get out of the SVC onto the FS9500.  Essentially, we create Zones to be able to map LUNs from the FS9500 into SVC.   Then, if we had for example a 200GB LUN on the SVC for an iSeries partition, we would create a 200GB on the FS9500, map it and discovered it under the SVC.   Then proceed to create a one to one, copy of the active 200GB LUN on the SVC onto the new 200GB mdisk discovered from the FS9500.

    # addvolumecopy -image mdisk_name_or_ID -pool svc_pool_name_or_ID svc_volume_name_or_ID

    We added a few and let them all synch over time.  Once fully synched, we waited for the proper maintenance windows (a monthly reboot due to patches or something similar).  With the server offline, we would then remove the volumecopy from SVC.

    # rmvolumecopy -pool 3 mdisk_name_or_ID

    From the FS9500, we unmapped the 200GB LUN from SVC, mapped it to the target server and restarted the server with the new LUNs.  Administrator had to do some clean ups due to "expected" LUID and the such but it all went fine.  Waited a little while and of course proceeded to clean up old zones, mappings, LUNs from SVC

    We had virtualization licensed already and it was ending shortly, but giving us enough time to complete our migration.  But I am sure IBM and/or vendor can help out as well in a migration situation as far as the licensing is concerned.  Reach out to them.

    Hope this helps and gives you some insights.

    Steph



    ------------------------------
    Stephan Durand
    Manager Information Technology – Storage & Backups, Disaster Recovery & Infra *NIX
    Rona Inc
    Boucherville QC
    5145995900
    ------------------------------



  • 3.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted Tue February 11, 2025 12:27 PM
    Edited by Sudhir BISHT Tue February 11, 2025 12:29 PM

    Hi Stephan.  Thanks for the great insights.  Also, will these commands need to be run on the FS7300 Service Assistant CLI?.  I mean are these FS7300 commands or SVC commands. Will these work for A9000R, like they worked for SVC.  SVC, as I understand is already a Virtualized box like EMC Vplex.   Thanks.





  • 4.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted Tue February 11, 2025 12:58 PM
    Edited by Sudhir BISHT Tue February 11, 2025 12:58 PM

    Hi. Ok. Since you had SVC, you can use 'Import in Image Mode and Export to Image Mode, as I read in the Migration manual.  In my case, we can only use 'Import in Image' mode.  But apart from that statement, in entire manual there is no information about what this 'Import in Image' mode is and how it is used.   Unbelievable.  Thanks




  • 5.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted Tue February 11, 2025 01:52 PM

    Hi.  Yeah, my example is a bit different, since I am technically in the same "Family" of code/commands/fonctions between SVC and FS9500.  You on the other end, with the A9000r, though it is similar, but has some discrepancies.   I meant it as an example, as a possible way to get the ball rolling in thinking of ways to possibly do it on your end, maybe the same way we did. 

    We did 90% of our migration without using the storage controllers to do them.  AIX, Oracle (ASM), Linux etc were easily done, live.  But we had to be creative with the few older physical boxes and iSeries (as400) ones, which needed a downtime.  It was a short downtime, since we were able to create that live "mirror" between the 2 arrays prior, and rebooted on the new only.

    steph



    ------------------------------
    Stephan Durand
    Manager Information Technology – Storage & Backups, Disaster Recovery & Infra *NIX
    Rona Inc
    Boucherville QC
    5145995900
    ------------------------------



  • 6.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted 29 days ago
    Edited by Sudhir BISHT 29 days ago

    Hi Stephan.  One question regarding the ASM Balancing and Migration you guys did for this.  After ASM Database migration, the disks were migrated to new disks successfully no longer belong to the ASM, but at the time of removing them from the lpar (server) it is not possible because some services are still running on these disks. This situation has happened to us in many activities of this type and the only solution is to execute a controlled restart of both nodes so that these tied disks are released.  Did you also face such scenario.  As, only solution seems to be controlled restart of the nodes involved.  Please advise.  Thanks.




  • 7.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted 28 days ago

    Hi.  Can't say we've had issues with the ASM migration.  It is a fairly simple task.  We added new LUNs to AIX partition (or Linux in some instances).  Discovered and configured them as needed and DBAs detected and added them into their ASM configuration.  Migration is done through a rebalance of the diskgroup across all available disks.  Once rebalancing is complete, DBA will "drop" the unrequired disks, then delete them from ASM.  AIX admin will then rmdev the device which will allow the storage admin to unmap the LUNs.

    When you say "Removing them from the lpar".  Are you at the DBA level, removing them from ASM?  At the AIX level (rmdev)? or at the VIO level using rmvdev?

    For information purposes.  Our partitions are running on redundant VIOs through NPIV, not using vdev.



    ------------------------------
    Stephan Durand
    Manager Information Technology – Storage & Backups, Disaster Recovery & Infra *NIX
    Rona Inc
    Boucherville QC
    5145995900
    ------------------------------



  • 8.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted 28 days ago

    Hi Stephan.  Thanks.  ASM folks successfully made the removal without any news and these disks no longer belong to the ASM. But when removing them from the lpar (server) (I have not yet confirmed, but this lpar part is always handled by AIX Team here) it is not possible because some services are still running on these disks. And they say - this type of situation the only solution is to execute a controlled reboot of both nodes so that these tied disks are freed.  Also, I believe our partitions are also running on redundant VIOs through NPIV, as we hear a lot anout VIOS/NPIV here.  Any way to find that out from Storage side commands or Fabric Zoning info. I do know we have NPIV zonings configurations. Thanks. 



    ------------------------------
    Bhuppi
    ------------------------------



  • 9.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted 26 days ago

    From the switch side, you could do a "switchshow | grep NPIV" to list the ports, where NPIV is indeed seen.  If you know the address of the physical or virtual port, you can do a "nodefind c0:50:76:0b:f9:2c:00:18" for example and it will come back to you with a result, where you can see the physical port number on the switch.  You can do a "portshow ##" and see all the NPIV ports seen on that physical adapter.  Now, I've seen and heard, that "normally" ports starting with C0:50 are indeed NPIV.  Not sure if that is ALWAYS 100% correct though.  In my case, it is.  But wonder if with different vendors, this would change?  Anyway.  From there, you can look at your zoning to see if you are indeed zoned using your NPIV ports.



    ------------------------------
    Stephan Durand
    Manager Information Technology – Storage & Backups, Disaster Recovery & Infra *NIX
    Rona Inc
    Boucherville QC
    5145995900
    ------------------------------



  • 10.  RE: IBM A9000R to FS7300 MIGRATION (STORAGE BASED)

    Posted 25 days ago
    Edited by Sudhir BISHT 25 days ago

    Hi Stephan.  Yes, we are indeed NPIV. 

     nodefind XXXDEBPXXX08_0138
    Local:
     Type Pid    COS     PortName                NodeName                 SCR
     N    02f182;      3;c0:50:76:0b:57:73:01:38;c0:50:76:0b:57:73:01:38; 0x00000003
        SCR: Fabric-Detected Nx-Port-Detected
        Fabric Port Name: 2e:79:88:94:71:92:ec:79
        Permanent Port Name: 10:00:00:10:9b:9b:12:66
        Device type: NPIV Unknown(initiator/target)
        Port Index: 377

    switchshow | grep -i 377
     377   12   41   02f180   id    N32       Online      FC  F-Port  1 N Port + 2 NPIV public

    portshow 12/41

    portWwn of device(s) connected:
            c0:50:76:0b:57:73:01:38    Device type: NPIV Unknown(initiator/target)
            10:00:00:10:9b:9b:12:66
            c0:50:76:0b:57:73:00:34   Device type: NPIV Unknown(initiator/target)
    Distance:  normal
    portSpeed: N32Gbps

    So, looks like "Our partitions are also running on redundant VIOs through NPIV, not using vdev"

    Also, from Power Community -  Few suggestions came:  a)  fuser /dev/<raw_device_name> ?   Check if the disks are being hold by Oracle processes or Aix.

    If they are Oracle processes, that means that the DB has not yet completely released them.

    b)  Try to unlock it before removing it: (in case the devices were locked)

    lkdev -l hdiskX -d

    Thanks