Broadberry Cyberstore running Windows Server 2008R2 Storageserver with Backup Exec 2010R3
The Cyberstore has a RAID array containing approximately 8TB of data, and a 150GB RAID0 system disk.
Connected to this is a HP DL380G6 server running vSphere 4 ESXi, and we have the correct agent, so the Cyberstore also backs up the six or so virtual machines.
Back up are to a Tandberg T24 autoloader fitted with an HP Ultrium LTO5 tape drive.
Full back-ups to tape of the entire Cyberstore and ESXi server and VMs takes about 37 hours, writes just under 9TB of data and uses six tapes.
I was tasked to carry out a full restore of a set of tapes, which is when I discover that whoever installed this setup hadn't configured IDR properly, so the .DR file was not accessible for the set of tapes chosen. I was able to create the IDR boot CD-ROM however.
I booted the IDR CD and selected a manual restore and pointed it at the first tape of the set. This was yesterday at approximately 3pm. It is now 10:30am the following day, and the system still has not completely read the first tape, and I estimate that it is perhaps 75% the way through.
The tapes are encrypted which might possibly slow things down, but am I really to expect my full restoration to possible take a couple of weeks!? Or is there something potentially wrong here?
I am going to terminate the restoration process shortly, because the system (C:) disk of the Cyberstore has been done. I have a theory that the system disk will contain a DR file for this tape set. It will also contain a proper install of BE2010, which might possibly perform the rest of the installation better. Does this sound like a reasonable thing to do?
I am going to change the procedures for backup to include having BE2010 write a copy of the DR file to a USB stick mounted on the Cyberstore. There will be a USB stick for each media set, which will travel with the set to the off-site back up safe. Does this sound a reasonable strategy too? I know in an ideal world there would be a replica server which is in sync with this one, but that is not the situation at the moment, and I have been given the go ahead to do that. In the meantime, I have this single dataserver with no redundancy, which I need to be able to restore, however long it takes.
Any guidance would be very much appreciated.
The IDR process automates the recovery of a server. The speed of recovery depends on the amount of data to be restored to recover the server. The IDR process is only faster in that there is no lag between the restoration of various components of the server.
If you do not have the .dr file, you can still manually recover your servers from your tapes. Just install the OS and BE on the new machine. Inventory and catalog your tapes and you can start your recovery.
Your idea of putting the .dr file onto a thumb drive is a good one, but you would have to ensure that the drive letter of all your thumb drives and the path to store the .dr file is the same for all your thumb drives. This is because you can only set one secondary IDR path.
I have managed a manual restore of the server C:, System State, and Shadow Copy Components. I was able to recover the DR file from this restoration, after I have manually recovered the main RAID array and the ESXi server and VMs, I will try using the DR file and see how this works. Especially how much quicker or slower it is, and how much is possible without manual intervention.
The next question I will need to answer for myself is whether or not it is possible to read a DR file from a USB memory stick from the IDR restoration boot CD-ROM.