HP ProLiant DL380 G5 DPSS, Руководство администратора

Страница: 65 / 168

Disk space for shadow copies can be allocated on either the same volume as the source fi les or a
different volume. There is a trade-off between ease of use and maintenance versus performance and
reliability that the system administrator must consider.
By keeping the shadow copy on the same volume, there is a potential gain in ease of setup and
maintenance; however, there may be a reduction in performance and reliability.
CAUTION:
If shadow copies are stored on the same volume as the user fi les, note that a burst of disk input/output
(I/O) can cause all shadow copies to be deleted. If the sudden deletion of shadow copies is
unacceptable to administrators or end users, it is best to use a separate volume on separate disks to
store shadow copies.
Determining creation frequency
The more frequently shadow copies are created, the more likely that end users will get the version that
they want. However, with a maximum of 64 shadow copies per volume, there is a trade-off between the
frequency of making shadow copies and the amount of time that the earlier fi les will be available.
By default, the storage server creates shadow copies at 0700 and 1200, Monday through Friday.
However, these settings are easily modi fi ed by the administrator so that the shadow copy schedule can
better accommodate end user needs.
Shadow copies and drive defragmentation
When running Disk Defragmenter on a volume with shadow copies activated, all or some of the shadow
copies may be lost, starting with the oldest shadow copies.
If defragmenting volumes on which shadow copies are enabled, use a cluster (or allocation unit) size
of 16 KB or larger. Using this allocation unit size reduces the number of copy outs occurring on the
snapshot. Otherwise, the number of changes caused by the defragmentation process can cause shadow
copies to be deleted faster than expected. Note, however, that NTFS compression is supported only
if the cluster size is 4 KB or smaller.
NOTE:
To check the cluster size of a volume, use the
fsutil fsinfo ntfsinfo
command. To change
the cluster size on a volume that contains data, back up the data on the volume, reformat it using
the new cluster size, and then restore the data.
Mounted drives
A mounted drive is a local volume attached to an empty folder (called a mount point) on an NTFS
volume. When enabling shadow copies on a volume that contains mounted drives, the mounted drives
are not included when shadow copies are taken. In addition, if a mounted drive is shared and shadow
copies are enabled on it, users cannot access the shadow copies if they traverse from the host volume
(where the mount point is stored) to the mounted drive.
For example, assume there is a folder
F:\data\users , and the Users folder is a mount point for G:\ .
If shadow copies are enabled on both F:\ and G:\ , F:\data is shared as \\server1\data , and
G:\data\users is shared as \\server1\users . In this example, users can access previous versions of
\\server1\data and \\server1\users but not \\server1\data\users .
Managing shadow copies
The vssadmin tool provides a command line capability to create, list, resize, and delete volume shadow
copies.
HP ProLiant DL380 G5 Storage Server 65