Migrating data with a smaller amount of free capacity Image mode and migrating existing data

This task provides step-by-step instructions for migrating existing data with a small amount of capacity.

Context

A small amount of free capacity is available. The RAID arrays containing the existing data will still contain the data after importing the arrays to the cluster.

Attention: Attention: This will result in an uneven distribution of data across the managed disks in the managed disk group. The severity of which will depend on how many managed disks there are in the managed disk group, initially, and how many of these have free capacity. For example, you have one managed disk in the destination managed disk group. You bring in image mode LUNs from an array on the storage subsystem. You migrate these LUNs to the destination managed disk group. These LUNs are now striped across the one managed disk. Now you add another LUN to the destination managed disk group. So, it has two managed disks in it now, but all of the data is on the first managed disk. No managed disks are in the second group and once again have very unbalanced storage. There will always a lot more data on the first few managed disks in this managed disk group than in the last managed disks that were added, and once again have very poorly balanced storage with 1 or 2 managed disks being extremely loaded. Some subset of data will have to be migrated from the overloaded managed disks to the under utilized ones.
Attention: Attention: This procedure may require subsequent migration of data within the managed disk group in order to even out the distribution of data across the managed disks in the group.
Steps

Perform the following steps for migrating existing data with a small amount of capacity:

  1. Select a managed disk group which contains enough free capacity to migrate all of the logical disks on the first array to be migrated into the cluster.
  2. Create an empty managed disk group this will temporarily contain the data being imported.
  3. Stop all I/O operations to the logical disks that are to be migrated first and un-map these disks from their hosts.
  4. Create an image mode virtual disk from the first managed disk that contains the data to be imported. To do this, perform the following steps:
    1. Map one logical disk from the storage subsystem to the SAN Volume Controller ports.
    2. Issue the svctask detectmdisk command on the cluster. The new managed disk that is found will correspond with the logical disk mapped in the previous step.
    3. Create an image mode virtual disk from this managed disk. Assign it to use the empty managed disk group just created.
    4. Repeat for all logical disks as required.
  5. If you have decided to continue the I/O operations while you migrate the data into SAN Volume Controller, map all the image mode virtual disks to the hosts using the SAN Volume Controller and continue accessing the data through the SAN Volume Controller.
  6. Migrate the data into the managed disk groups you created in step 1. To do this, perform the following:
    1. Select the first image mode virtual disk to be migrated.
    2. Migrate this virtual disk from its current managed disk group into one of the managed disk groups created in step 1 above. This will migrate all the data from the logical disk into the new free space.
    3. When this completes, select the next image mode virtual disk and repeat the previous step.
  7. The RAID array that contains the logical disks can now be re-configured and added to the managed disk group selected in step 1. To do this, perform the following steps:
    1. Remove the managed disks from the temporary managed disk group.
    2. At the storage subsystem, the logical disks that have been migrated should be unmapped from the SAN Volume Controller cluster and deleted from the array (if more than one existed).
    3. Assuming the array meets the guidelines, a single logical disk should be created using the entire array capacity.
    4. This new logical disk can be mapped to the SAN Volume Controller ports.
    5. Issue the svctask detectmdisk command on the cluster. The new managed disk that is found will correspond with the new logical disk created.
    6. Add this managed disk to the managed disk group selected in step 1.
  8. Repeat steps 3 through 7 for the next array.
Related topics Image mode and migrating existing data

(C) Copyright International Business Machines Corporation 2003, 2004