weekly make_net_recovery?

[SDS FDX]

Using my ignite server, I'm looking to create daily or weekly system images of our 11.11 systems, using the make_net_recovery utility. I have two options: (1)Using make_net_recovery, create the system image with the minimal system files in rootdg, which uses about 450Mb of disk space. Once system is recovered using the system image, restore the remaining system and data files using Netbackup. (2)Using make_net_recovery, include all of rootdg in the system image, which uses about 1.2GB of disk space. Once system is recovered using the system image, restore only the data files outside of rootdg using Netbackup. In option (1) I'm concerned about restoring O/S files on a running system. In option (2)disk space to store the images is an obvious concern, when we begin receiving more HP servers. Any thoughts????

[A. Clay Stephenson]

All I ever concern myself with is vg00 with respect to make_tape_recovery or make_net_recovery. Everything else is done via standard backup/restore (Data Protector in my case), NetBackup in your case. I would use option 2) because A) it's much easier (and safer) to get everything in the VG B) you avoid replacing shared libraries and executables 'on the fly'.

Disk is cheap and you really need to keep only 2 versions of each server's Ignite image or 1 if you backup these images to tape. I find that weekly Ignites are more than sufficient because very little in / actually changes once the system stabilizes. The obvious exception to the weekly is just before patching.

Let me give you a plan B to think about. Add a "lifeboat" disk to each system. This is simply a raw disk copy of your boot disk(s) made each week while the system is relatively quiet. If something terrible happens, you simply powerdown, yank the boot disk(s) and insert the lifeboat(s) in the boot disk slot(s) and you are back in actiuon in a fraction of the time an Ignite takes. You should still do Ignites for Disaster Recovery.

[SDS FDX]

I understand that it's pretty easy to backup everything under rootdg, but I have to take into consideration that each full system image (2 copies per server) at about 1.5Gb * number of systems, can take up a bunch of diskspace. I guess, if I can get a low end SCSI attached array, disk space won't be an issue. The idea of restoring remaining system files while the system is up and running, does scare me. Thx for your comments! Anyone else have any ideas????

[Steven E. Protter]

Shalom Richard,

You can store the images anywhere. Any NFS mount point will do. A used up old Linux box with an NFS server will work(Probably).

You can build an old Ignite server from older end of life parts. You can write the images weekly and then archive them to tape for later use.

Lots of options, just need to think outside the box.

I recommend a weekly make_tape_recovery after some testing to see what storage solutions work for you.

SEP