Replacing All E-Ddms At The Same Time - IBM System Storage DS5020 Installation, User's, And Maintenance Manual

After replacing all the E-DDMs, you must reconfigure the new E-DDMs and
restore data from backup. See the procedure in "Replacing all E-DDMs at the
same time."
This is the safest way to exchange E-DDMs without losing data. However, this
method might take a long time to complete because of the backup,
reconfiguration, and restoration processes. In addition, other users are not able
to use the storage subsystem (or any storage expansion enclosures attached to
the storage subsystem) until you finish the procedure. You must use this method
on RAID 0 logical drives.
v Replacing the E-DDMs one at a time
In this method, you manually fail each E-DDM, replace it, and wait for the system
to restore data to the new E-DDM before installing the next E-DDM. After you
install the new E-DDMs, you can configure them to make the additional drive
space available. See the procedure in "Replacing the E-DDMs one at a time" on
page 138.
Using this method, you can replace the E-DDMs while the storage expansion
enclosures and the DS5020 are running, eliminating the down time that is
required if you replace all the E-DDMs at once. However, this method is more
risky because you can lose data if the drive restoration or storage subsystem
reconfiguration process fails. In addition, the reconstruction process might take a
long time. This method works only on redundant logical drives (RAID 1, 3, or 5).
You cannot use this method with any drives containing RAID 0 logical drives.
Consider backing up your data if you use this method. This safeguards your data
if the restoration and reconfiguration process fails or the new drive malfunctions.
The method you use depends on the following considerations:
v Which method most closely matches the drive upgrade procedure in the
operating-system or the storage-management software documentation.
v Which RAID level is used on the affected drives. (RAID 0 requires you to replace
all the E-DDMs at the same time.)
v How much downtime is acceptable as you swap the E-DDMs.
v The number of E-DDMs in an array. Replacing E-DDMs one at a time is better
suited for arrays consisting of 3 to 5 E-DDMs. If you have more than 10 E-DDMs
or arrays that are comprised of high capacity SATA E-DDMs, consider replacing
all E-DDMs at the same time.
v How much risk of losing data is acceptable. Because the array will be in a
degraded state during the RAID array reconstruction and copyback process as
result of replacing a drive in the array, any new E-DDM failure will cause the
array to fail (causing a loss of data availability and even a loss of data). The
duration of the reconstruction and copyback process can be quite long,
depending on the size of the RAID array.
v How extensive will the data be changed while the array is in a degraded state
during the RAID array reconstruction and copyback process as result of replacing
an E-DDM in the array. The more extensive the data changes, the more work
that will need to be performed to restore the data in the event that the array
failed because of an addition E-DDM failed while the array is in degraded state.

Replacing all E-DDMs at the same time

Use this procedure to replace all E-DDMs at the same time. You must use this
method if you are upgrading E-DDMs containing RAID 0 logical drives. All the data
currently on the E-DDMs is lost when you replace the E-DDMs; therefore, you must
back up all data that is currently on the E-DDMs. This procedure also requires you
Chapter 5. Replacing components
135