TR Forums

I'm currently in the process of backing up multiple PCs, Linux and Windows, and there is a serious PITA with Windows boxes.

I have a lot of data I need to back up and merge, some of which is being dragged from Win 98 or something era, via "My Documents" upgrades to the next version, some of which was copied by brute force when updating HDDs, and some of the data is produced by apps that have been used relatively recently.

So there is some data under "Program Files (x86)", some of it is in "My Documents" or "Documents", something seems to be under "%userprofile%\AppData\Local\VirtualStore", something goes into Public profiles, and they have some dummy sample data as well, and ton of other places. With Win7 it also seems that there is a ton of junction points, so robocopying works with extra switches, but I'm not sure it does what I need. And it definitely has all the extra crap like profile specific temp files and stuff which is heavy and unneeded. Like NTUSER.DAT and registry hives or something, if I understand correctly.

Does anyone know a good checklist by which I could avoid copying whole "My Documents" folders, but not leave anything behind? As of now it's nightmare, I know about 10 places to look, but I'm really not sure it's everything.

Doing anything in Windows seems awful after getting used to Linux.

It's a nightmare; Microsoft is partly to blame by constantly changing their mind about the location of the various home folders, but your real culprit is the software writers who dump save files and progam information all over the place.

You'll never find it all, because half the time the people who wrote the software in question don't even know where it is (*cough* Adobe, Symantec).

Yeah, the only way to be /sure/ is to backup the whole drive.

Probably easiest to have a Windows fileshare set up and have people use the Win7 backup utility to make a full system backup + incrementals. Make sure you get a system image. Natürlich, this'll work better with GigE all around.

If your concern is getting all the data backed up just throw some storage at it and do like bthylafh said. Use the built in Windows Backup and create a system image. Any time you spend trying to track down and catalog user data that has gone through a series of (it sounds like) poorly executed upgrades will massively offset the price difference of a 500 gig to a 2 terabyte drive. Get your backups working correctly and then (if you have the time/desire) go back and clean up the user data.

Backups are always the priority.

Mind you, NTUSER.DAT is not something superfluous, it's the most important file in a user profile which stores the content of the HKEY_CURRENT_USER registry hive.

However, this file, and probably several others, are open for writing at all times when the user is logged in. Which means ... no matter what utility you use to copy them, you can get corrupt copies as a result.

To back up my profile, I first log off, and log on as administrator. Then I copy my profile to the backup drive. At this stage I skip some large files that I've previously identified as unnecessary. The Google Earth cache, for example. That's it. Pretty paranoid, heh?

Volume Shadow Copies are your friend and mine. That's how you get good copies of in-use files on Windows.

Backing up Windows hosts is a right royal pain in the backside. Believe me, I know - backup and recovery is what I do for a living (if you care: currently TSM, soon to add Avamar to the list.) It doesn't matter whether Windows is 10%, 25%, 50%, 90%, or 100% of the environment, I guarantee you: well over 95% of my problems at work come from Windows hosts. Yes, even if they're only 10% of the environment. Unix (AIX, Linux, HP-UX, Solaris) just keeps quietly working; Windows, it seems like I have to hand hold it every step of the way.

First thing first: are you looking to back it up so you can do a full system restore, or just so you can restore documents after you've rebuilt the box from the ground up? If the first, you need to make sure you backup the system state - Windows 2003 could do this out of the box with NT Backup, but I'm not sure if that tool is included in Windows 7. ... *checks* looks like it is: control panel, backup and restore. Mind you, I can't check to see if it does what it should do - I'm at work, don't have an external hard drive to play around with. Just remember that, when you restore the system state, you need the hardware on the system you're restoring to to be identical to the hardware of the original box. Sucky, yes, but that's the way it is.

Then you need to backup the data. Suck it up: buy enough capacity to backup all the data off the entire hard drive. Yes, it's annoying to have to back everything up. Bluntly, it's a lot easier to delete data that didn't need to be backed up after the fact than it is to backup data that did, but wasn't.

Another option would be to install Cygwin, and do a full disk MD5 or SHA1 hash. Look through the list for duplicates; do a file compare to make sure that the hash wasn't lying to you, and then you can delete duplicate files.

But don't go trying to trim down the backup to just what you think you need. The absolute most you should do is exclude directories that you know you don't need. The former guarantees that you'll miss something important. The latter means that you might be backing up more than you "should", but at least you won't miss something you do need. If in doubt, back it up. (for example: I might use the Steam backup facility to stash a copy of my games somewhere, and not worry about backing up the entire Steamapps directory. I lose out on save games, but I'm usually not fazed by that; at least I don't have to re-download all that application data.)

And don't forget to do restore tests. A backup that hasn't been tested might as well have been sent to /dev/null (or the Windows equivalent). There are no guarantees, even with a restore test, but I have far more faith in backups that have been tested than in ones that haven't been.

There are some pretty good dupe file finders for Windows. This one is free:
http://www.auslogics.com/en/software/du ... le-finder/

and this one is $30 per user:
http://www.scootersoftware.com/moreinfo.php

Using sysprep and ImageX can allow for a file based backup of Windows that is not tied to any specific hardware, mountable through ImageX or DISM, or you can convert the .WIM file into a .VHD and mount that as a drive in Windows.

USMT and Windows Easy Transfer Tool are the utilities meant to be used to migrate data off a system and then back onto a new system. USMT through the hardlink migration store can even do this without you needing to move that data off the disk.

As for where the data is hiding... that depends on what applications you use. Most of the symbolic links Windows uses is simply for compatibilty for old installer programs that though that hard coded paths was a good idea.

I'd note that Windows Backup and Restore can also generate backups as a .vhd.

Volume Shadow Copy cannot prevent file corruption in every case, as explained here:

http://technet.microsoft.com/en-us/library/cc785914%28v=ws.10%29.aspx

... the shadow copy will still occur and all of the data, in whatever form it is in at the time of the copy, will be included in the shadow copy. This means that there might be inconsistent data that is now contained in the shadow copy. This data inconsistency is caused by incomplete writes, data buffered in the application that is not written, or open files that are in the middle of a write operation.

Even though a corruption is very unlikely, I'd rather copy a system volume when it's not mounted, if this is an option.