Very Computer

or like
Disk1           Disk 2          Disk 3          Disk4
1               3               5               7
2               4               6               8

Also, does this qualify as a disk array?

Thanks,

- Dc.

Howdy,
I recommend you take a look at
http://www.redbooks.ibm.com/abstracts/sg245433.html

It's a very good redbook on the LVM

Daniel

|> Hi.. I am basically an Oracle DBA person who dabbles in sysadmin only
|> so much.  I was hoping I could tap this group for a question.  
|> I am told I have Raid capable SSA controlers, but currently my disks
|> are not configured for RAID, but instead have some logical volumes
|> spread across several physical disks at the PP level.  I was hoping
|> someone would know whether this would imply that PP's are round
|> robined across those physical disks, or just clumped one group at a
|> time.  I am interested as to whether I get any parallel read ability
|> with a datafile in one of these logical volumes.  I guess one
|> possibility is that both are possible, in which case it would be
|> helpful to know how that is done, and more importantly, how I can tell
|> wich was done.   In other words, does this technique imply a sort of
|> primitive striping or not? If I have 4 disks on the lv, will it go
|> like
|> Disk 1                Disk 2              Disk 3                Disk4
|> 1         2               3               4
|> 5         6               7               8
|>
|> or like
|> Disk1             Disk 2          Disk 3          Disk4
|> 1         3               5               7
|> 2         4               6               8
|>
|> Also, does this qualify as a disk array?
|>
|> Thanks,
|>
|> - Dc.

A quick primer on lv allocation:

The base command for creating lv's is mklv.  The command uses allocp to allocate
partitions.  There are two options for partition allocation when multiple
drives are used - 'minimum and maximum.  With minimum, the number of requested
partitions are allocated from the first drive until either the request is
fulfilled or the drive is full.  If the drive fills, then the next drive in the
list is used.  This goes on until the request is fulfilled.  If there are not
enough free partitions on the drives, the allocp call fails.
With a policy of maximum, the partitions are allocated round-robin, starting with
the first drive on the list and continuing onto the next drive and the next
drive, coming back to the first drive.  Whenever a  drive fills up, it drops out
of the loop.  For things done this way, you get data placement with a granularity
of a PP.  For striping, you get the same style of allocation as maximum, but the
data placement granularity is at the stripe size setting (much smaller than PP
size).

To see what you have you do 'lslv <lv-name>' and look for "INTER-POLICY:" value
of maximum or minimum.  If the lv was striped, the stripe width (number of pv's
used for allocation) and stripe size will also be displayed.  To see the logical
to physical relation of an lv, use the command 'lslv -m <lv-name>'.  This will
show the layout for lv's that are mirrored as well.

Hope this helps.

--

Senior Software Engineer- Worldwide OEM Marketing - Consulting and Tech Support
Enterprise Server Group,  IBM-Austin
<All standard disclaimers apply. Non-standard disclaimers do too.>
--

Senior Software Engineer- Worldwide OEM Marketing - Consulting and Tech Support
Enterprise Server Group,  IBM-Austin
<All standard disclaimers apply. Non-standard disclaimers do too.>

Jaime,

Urban

This e-mail was composed of 100% recycled bits.

|> This is a multi-part message in MIME format.
|> --------------D963E033998B927E15426726
|> Content-Type: text/plain; charset=us-ascii
|> Content-Transfer-Encoding: 7bit
|>
|> Jaime,
|>
|>

|> > |> Hi.. I am basically an Oracle DBA person who dabbles in sysadmin only
|> > |> so much.  I was hoping I could tap this group for a question.
|> > |> I am told I have Raid capable SSA controlers, but currently my disks
|> > |> are not configured for RAID, but instead have some logical volumes
|> > |> spread across several physical disks at the PP level.  I was hoping
|> > |> someone would know whether this would imply that PP's are round
|> > |> robined across those physical disks, or just clumped one group at a
|> > |> time.  I am interested as to whether I get any parallel read ability
|> > |> with a datafile in one of these logical volumes.  I guess one
|> > |> possibility is that both are possible, in which case it would be
|> > |> helpful to know how that is done, and more importantly, how I can tell
|> > |> wich was done.   In other words, does this technique imply a sort of
|> > |> primitive striping or not? If I have 4 disks on the lv, will it go
|> > |> like
|> > |> Disk 1                Disk 2              Disk 3                Disk4
|> > |> 1            2               3               4
|> > |> 5            6               7               8
|> > |>
|> > |> or like
|> > |> Disk1                Disk 2          Disk 3          Disk4
|> > |> 1            3               5               7
|> > |> 2            4               6               8
|> > |>
|> > |> Also, does this qualify as a disk array?
|> > |>
|> > |> Thanks,
|> > |>
|> > |> - Dc.
|> >
|> > A quick primer on lv allocation:
|> >
|> > The base command for creating lv's is mklv.  The command uses allocp to allocate
|> > partitions.  There are two options for partition allocation when multiple
|> > drives are used - 'minimum and maximum.  With minimum, the number of requested
|> > partitions are allocated from the first drive until either the request is
|> > fulfilled or the drive is full.  If the drive fills, then the next drive in the
|> > list is used.  This goes on until the request is fulfilled.  If there are not
|> > enough free partitions on the drives, the allocp call fails.
|> > With a policy of maximum, the partitions are allocated round-robin, starting with
|> > the first drive on the list and continuing onto the next drive and the next
|> > drive, coming back to the first drive.  Whenever a  drive fills up, it drops out
|> > of the loop.  For things done this way, you get data placement with a granularity
|> > of a PP.  For striping, you get the same style of allocation as maximum, but the
|> > data placement granularity is at the stripe size setting (much smaller than PP
|> > size).
|> >
|> > To see what you have you do 'lslv <lv-name>' and look for "INTER-POLICY:" value
|> > of maximum or minimum.  If the lv was striped, the stripe width (number of pv's
|> > used for allocation) and stripe size will also be displayed.  To see the logical
|> > to physical relation of an lv, use the command 'lslv -m <lv-name>'.  This will
|> > show the layout for lv's that are mirrored as well.
|> >
|> > Hope this helps.
|> >
|>
|> So I have to ask the follow on question, if you change the INTER-POLICY to maximum and
|> reorgvg the volume group, will you get striping at the PP size level, or does it ignore
|> this because the PPs were already allocated. I always wanted a complete understanding of
|> this, but no space to test.
|>
|> Urban

Yes.  If you set the INTER-POLICY from min to max ( use smit chlv) and then do a
reorgvg you will get a change in the placement of the PPs across multiple drives.

It is important to note that the RELOCATABLE flag must be set for the lv.

The reorgvg command is a shell script in /usr/sbin.

Jaime

--

Senior Software Engineer- Worldwide OEM Marketing - Consulting and Tech Support
Enterprise Server Group,  IBM-Austin
<All standard disclaimers apply. Non-standard disclaimers do too.>
--

Senior Software Engineer- Worldwide OEM Marketing - Consulting and Tech Support
Enterprise Server Group,  IBM-Austin
<All standard disclaimers apply. Non-standard disclaimers do too.>