Systems being inherrited is nothing new. Whether it is from IBM, a business partner, or just whoever else was before you. History is dependent on goals and/or problems as people saw them then.

Before I can comment (as you wisely said no comment as well) I need some feedback on current plans and or problems. I still do not understand your goals - although I am beginning to understand some of your resources (i.e. a p570 (power6), or a p5-570 (power5)) and some SAN storage.

Without worrying about impact on current configuration: what are your goals and/or problems. In other words, what do want to achieve as system administrator or application administrator/owner.

Otherwise I can only list different methods to configure a POWER system but never really address your problem.

Basically, regardless of your choice of LVM mirroring, or MPIO to SAN based RAID/mirored storage, the first thing you will need is a dual VIOS setup.

As you are using only one VIO Server atm - at least that is how I understand your reply - you do not need two vhost/vclient adapter pairs for your storage. The LUN values are only unique to one vhostX adapter - as they are LUNs of that adapter, not of the VIO Server; not for the client.

As the material is a bit complex - I normally teach this as part of a 4 day course for VIO Best Practices, I shall not attempt to give a quick and dirty (or quick and easy) answer here.

This is a topic I have been working on. Graphics is hard - but I am getting better. And as those start shaping up I'll get the articles finished.

1) Where is remote storage coming from? SAN (LUN), physical disks, or as logical volumes?
2) How many VIO partitions per managed system?

Since you have two vscsi adapters (one on C4, one on C5) the LUN can be repeated.

This also means you have two vhost adapters on the VIO server - or - you are using dual VIO servers and lvm mirroring (perhaps).

If you have only one VIO server, you do not need a seperate VSCSI server/client adapter pair for each disk.

Planning ahead helps a lot. When you create the vtscsi object you can add the -dev argument to give it a name that is meaningful to you: e.g. -dev lp10_hd3 (for lpar10, hdisk3).

However, if you are like me - and learn handy things like this after a large install - the commands you want to look at on the VIOS is lsmap and on the AIX partition is lscfg.

On my current system I have given the LP a descriptive name. Notice the LUN value. Also note the Client Partition ID.

$ lsmap -vadapter vhost0
SVSA            Physloc                                      Client Partition ID
vhost0          U9110.51A.0639B9D-V1-C11                     0x00000002

VTD                   vtopt0
Status                Available
LUN                   0x8200000000000000
Backing device        cd0
Physloc               U788C.001.AAA4406-P1-D5

VTD                   vtscsi0
Status                Available
LUN                   0x8100000000000000
Backing device        lp2vd1
Physloc               

VTD                   vtscsi1
Status                Available
LUN                   0x8300000000000000
Backing device        lp2cd2
Physloc               


Connect to the Partition with ID #2 and run the following command:

# lscfg -v | grep V2 | grep L8
  hdisk1           U9110.51A.0639B9D-V2-C2-T1-L830000000000  Virtual SCSI Disk Drive
  cd0              U9110.51A.0639B9D-V2-C2-T1-L820000000000  Virtual SCSI Optical Served by VIO Server
  hdisk0           U9110.51A.0639B9D-V2-C2-T1-L810000000000  Virtual SCSI Disk Drive


The grep for V2 is because the partition is #2, so it's physical location for virtual devices will always include V2-C (with the virtual slot number following the -C. In this case C2.). The added L8 is just to make the list smaller. You may actually want to skip that to see everything attached via the virtual i/o planer.

All that is left now ios to match up the LUN values L8* and you have your disks.

Hope this answers your question!

Michael