In article < XXXX@XXXXX.COM >,
mksysb on AIX or Ignite on HP/UX will be your closest solution to
restoring fully from tape. I haven't used Netbackup's Bare Metal
Restore (BMR). If you're going to evaluate it, I'd ask for pre-sales
support to restore the following types of systems:
- failed client system
- failed client system w/ a volume manager and logical volumes
- failed server system
- failed server system w/ a volume manager and logical volumes
Supposedly, Netbackup is a modified version of GNU tar. If that can be
restored without restoring the server, great. But what about the slices
or logical volumes? mksysb saves this information. Does Netbackup?
Before BMR, the conventional disaster recovery approach was to backup
data with whatever's you enterprise solution (Networker, Netbackup,
whatever), have a ufsdump of the system disk, Solaris Install CDs,
Backup software media, and a hardcopy of the layout of the filesystems
and disks ready at hand.
You'd install Solaris from CDs using the slice layout from hardcopy,
restore the system disk from ufsdump tape, configure the volume manager
and logical volumes, install the 3rd-party backup software and configure
it either as a server or client, and finally restore data from the
My concern is the 24-hour deadline. Depending how much data you have to
restore and the number of systems you have to restore, you may not make
it. If you have 20 Solaris systems, do they all have tape drives and do
their backups local? How long does a backup take? I'd double or triple
that time for a restore. If you do a dry run you'll get a better feel
for how long this will take.
It will lead some hard choices for management on what they're will to
spend to meet the 24-hour metric, if it indeed can even be met.
DeeDee, don't press that button! DeeDee! NO! Dee...