offsite system backup for HP Proliant ML150G5 server

[newind]

I'm consulting a background scheme for our HP server.

BACKGROUND
I have a HP Proliant ML150G5 server. I've configured it as a file server using Suse Linux Enterprise Server version 10.

The server has two 146GB SAS Hotplug Hard Drive configured as RAID 1+0. Since there are only 2 harddrives, correct me if I'm wrong, in effect, this is just RAID 1 (mirroring).
The server is working well so far and we're happy with it.

QUESTION
I would like to have an offsite system backup, so that in case something goes wrong with the OS (say, I accidentally delete some system files), I could bring up the system without having to re-construct all the hardware and software settings.

Assumption: Apart from the system backup, I will keep a separate data back up in another disk or machine.

I see 3 possibilities here.

OPTION 1
Given
-2 plugged SAS harddrives labeled A & B (current setup)

Unplug drive B and keep it as an offsite system backup. (Since my RAID setting is mirroring, I suppose I can do this without shutting down the server.)

If something goes wrong with Drive A (hardware error) or something goes wrong with the OS, I can still rebuild the system by unplugging Drive A then plugging in Drive B (to the same bay from which I got it from).

Disadvantages of this scheme:
1. There will be an additional downtime for getting Drive B into the office.
2. Update of data in Drive B (from the data backup )may take some time since it will be quite old.


OPTION 2
Buy another harddrive (of the same specs as the existing ones). Use this spare harddrive to swap with one of the two currently-installed harddrives every month.

Thus:

Given
-2 SAS harddrives labeled A & B
-1 extra SAS hardrive labeled C

Current Config
-Drive A is plugged in Bay 3
-Drive B is plugged in Bay 4
-Drive C is free, not plugged to server

After 1 month
-Hotswap Drives B and C (Uplug Drive B & replace it with Drive C in Bay 4)
-keep Drive B as offsite backup

After another month
-Hotswap Drives C and B (Uplug Drive C & replace it with Drive B in Bay 4)
-Keep Drive C as offsite backup

After another month
-Hotswap Drives B and C (Uplug Drive B & replace it with Drive C in Bay 4)
-keep Drive B as offsite backup

and so on...

In other words,
Drive A stays in Bay 3 while Drives B & C alternate occupying Bay 4, the one which is not plugged is kept as an offsite system and data backup.

Advantages of this scheme:
1. If one of the plugged harddrives fail, the system is still up (no downtime)
2. If both currently plugged drives fail, the system can be rebuilt from the offsite backup
3. since the offsite backup will be at most 1 month old, the update of data (from the data backup) will be much faster than in Option 1.

OPTION 3
Buy a 3rd harddrive, plug it in an empty bay, make it mirror the 2 existing drives, then unplug it to keep it as an offsite backup.
I'm not sure if this is possible with the RAID 1+0 setup (or for other RAID setups for that matter). If it's possible, i would appreciate getting instructions on how to set it up, given that I already have 2 harddrives setup as RAID 1+0.

Advantages of this scheme:
Same as in Option 2 plus: Drives A & B remain in tact in the server

BOTTOM LINE
I wish to know which of the 3 options above would be best. Or perhaps you can suggest a better solution. Just to repeat myself: What I basically want is an offsite system backup such that if, something goes wrong with the OS (or the haddrive) I can easily reconstruct the server.

In Windows desktop machines, I easily do this using a disk cloning/imaging. I'm basically trying to look for its equivalent in a RAID-setup, linux-installed, HP server.

Would appreciate your help.

Thanks in advaced.

[TTr]

Option3 can not occur. You can not have 3-way mirroring.

Option1 is not a good idea either. If you pull drive B, the raid controller will be in a downgraded stated and may be giving out all kinds of errors.

Option2 is a good idea except with one caveat. It will work fine the first month but the second month the spare disk already contains raid information on it, so when you plug it in, the raid controller will read it as a raid disk and will not sync drive A to it. You have to find a way to erase the disk before puting in the server.

Whatever you do you have to test your method and verify it will work, you can't just put a disk on the shelf and rely on that.

[marcus1234]

Antonio you seem to going along way about this the prefered method is a offline tape backup with feks.. a dat tape drive .....

and when data is written it is also checked ,

if data is valuable it is worth investing in a tape drive

if not so valuable just store it on an external drive they are cheap as chips and fast..enough..


you can also store the ghost image of the OS on the same drive too incase of disaster..
and restore in minutes

enjoy:)

This is how we thank each other in the forum
to show our appreciation to all those who post replies..

http://forums11.itrc.hp.com/service/forums/helptips.do?#33

Enjoy:)


You must be logged into the forum to assign points.

[newind]

TTr, thanks for the feed back. Any leads on how to resolve your caveat below?
=========
Option2 is a good idea except with one caveat. It will work fine the first month but the second month the spare disk already contains raid information on it, so when you plug it in, the raid controller will read it as a raid disk and will not sync drive A to it. You have to find a way to erase the disk before puting in the server.

[newind]

TTr,

I got the following reply from another forum about your comment on Option 2. Would appreciate receiving your thoughts on it.
===========
That is true if you plug in the hdd while server is offline (server power off), so to prevent this issue, you MUST plug in the hdd while the server is still online / powered ON then HP Smart Array (Raid controller) will detect it as replacement disk and automaticly start the rebuild process. The only short coming is, every time you change the hdd, the process rebuilding will taking place and take from minutes to hours to finish depend on your hdd size, so while the rebuild time your server does not have fault tolerance on hdd until the rebuilding process completed on that just plugged in hdd.

[newind]

mark1234

Sorry, I don't know what you mean by feks.

Do you think I can use Ghost to image the
RAID 1-configured SAS drives to an external hdd?

[TTr · Accepted solution]

I can not confirm the offline/online cases. I would expect that the controller will read the inserted disk the same way whether it is inline or offline. It will see the previous raid configuration on it and would treat it as a raid disk and not as an unused disk to start the rebuild.

The rebuild process can take several hours.

As I said, you can not create a "recovery strategy" on paper or as a concept only, you need to put it to the test before you can rely on it.

[marcus1234]

ok feks is abbreviation for example



ghost should do the job give it a try you will know in minutes..then it will be tried and tested which needs to be done for all backups..

[newind]

Thank you TTr and mark1234 for the help.

[TTr]

This URL has some really good information for splitting and combining mirrors.

http://bizsupport1.austin.hp.com/bc/docs/support/SupportManual/c00378986/c00378986.pdf

[newind]

Wow TTr! That link contains the information that I was exactly looking for. Millions of thanks!