Unsolved

This post is more than 5 years old

4 Posts

8720

February 26th, 2008 22:00

use of hotplug disc as backup medium?

Hello all,

I am scrambling to get a replacement backup system up after an ait-tapestreamer failed.
There is a NAS mirroring system in place, but the boss wants a copy he can take home every day.
The server is a poweredge 2800 and scsi discs raid 1 and 5. the backup runs at night and is about 100Gb.

As delivery of the new AIT-tapestreamer got unexpectedly delayed for another 2 weeks, I need something now.
As a stopgap measure, I'm putting a copy on an external usb disc. That'll appease the boss a little.

But it got me thinking, there are unused slots in the disc array/assembly.
I suppose I can add an extra disc in such as slot as a standalone volume and backup to that disc/volume.

Questions arise:
1) will the system allow for the disc to be replaced with an identical one...
     - without raising system alarms or rebuild request ?
     - while keeping the same windows drive letter?
  
2) if system alarms are raised, can they avoided by batch routines?
      the script could, at the end of a backup cycle, deactivate the disc and reactivate at the start of a cycle.

3) Can the backplane withstand frequent replacements of the disc trays ?

btw: know of any NAS with such a feature (rotation of discs)?

best regards
glenn


1.2K Posts

February 26th, 2008 23:00

As soon as a disk is removed (or fails) the system will detect this and report an error accordingly, there is no way around this. If the disk is apart of a redundant array (Raid 1,5 etc...) it will want to rebuild it when another disk is inserted. If the disk is a Raid 0 (which is what you will have to create if you want the disk visible in the OS), it will completely fail and you will have to recreate another Raid 0 container everytime it fails/is removed.

You could theoretically create a Raid 1 container (using two disks, obviously) and backup to this. Then at the end of the day, remove this (by taking it offline manually through OMSA or array manager, not just yanking the drive) and replace with another. This will then rebuild from the other disk in the mirror, ready to go the next evening. This whole process will require at least three disks in total.

But then you have problems when you put the disk in that has already been apart of the array, in that the system will probably see that it has been an array member and won't automatically rebuild to it. It would probably need the config erased from the disk.

 

Do yourself a favour, stick with the external USB disk.

36 Posts

February 28th, 2008 00:00

i think, it's visible, but not good idea.

you can configure new disk as non RAID partition.

with additional software for backup, you can sync data within the disc.

as it's a hotswap drive, you can put and pull, by force online and force offline the drive.

 

but, now external USB,Firewire, or NAS are very very cheap.

you can get 500GB of external USB disk below $200, and 160GB below $100.

why to bother having unprotected case internal drive, if you can take external HDD with rugged case?

 

may be the problem about speed. but if you set backup incremental atau differential, it should no problem. only changed data will be processed to backup.

4 Posts

February 28th, 2008 07:00

Yesterday night, I ran a complete backup to an external USB2 disc.
To my surprise, the performance was better than the scsi system.
Scsi  resulting backup size 80 GB - hardware compressed - time 5 hour (including verify)
USB  resulting backup size 120 GB - no compression - time 4 hour (including verify)

Ok, the usb didn't have the compression to do but a device costing $100 should not embarass a $2000 one.
at the moment,the current usb setup is a bit amateurish, but Backup to Disc is clearly the way to go.
inexpensive hardware, readily available, heck the usb disc cost about the same as one tape!

I discarded the idea of using the discslots of the raid system: It'll mostly likely not work as I want.
plus the hardware imho will not survive many bungled disc replacement attempts.

best regards
glenn

720 Posts

February 29th, 2008 18:00

Glenn dm,

 

  I agree with comment about H/W not surviving long, I think the connectors have a MTBF of 200 insertion /removal cycles, or 6 months to a year with daily cycles. Two hundred cycles is plenty for the intended usage of only the replacement of a failed drive or for upgrades. Most servers do not see 10 cycles in 5 years.

 

  Dell is offering ruggedized disk backup systems in both internal and external versions. The hard drives are treated just like a tape cartridge, and the chassis provides the mounting and  if external, the power and cabling connections so you can leave it connected all the time, but still secure the media or send it to offsite storage. I think they are called RD1000 or something similar. They use SATA drives, so the drives are about the same as the external USB.

 

Regards,

warwizard

6 Operator

 • 

1.8K Posts

February 29th, 2008 20:00

 

Drive racks, rated for 25k insertions, IDE,SATA,SCSI, SAS....

 

http://www.cru-dataport.com/htmldocs/products/dataport25/index.html

 

 

No Events found!

Top