Jeff W said:
HI VanguardX - thanks for the detailed response
1) Not interested in backing up to recordable media - only
disk-to-disk.
The NT Backup utility will backup to file. I just don't remember if it
supports external drives, like connected to a USB or firewire port.
2) What the heck is "EFS-protected" files - how do I
know if I have any?
EFS = Encrypted File System
It lets you encrypt files (or folders and any files put into them).
Like security certificates that you can get for digitally signing and/or
encrypting your e-mail, enabling EFS will create a security certificate.
You must remember to export the EFS certificate and secure the floppy
(or whatever you saved it on) because you might need it later, like for
a logical file restore (hard drives do die). If you don't have that
security certificate, and if no other account ever got designated as a
file recovery agent (Windows XP requires you to define one if you want
one whereas Windows 2000 automatically included the local Administrator
account), even you won't be able to read your EFS-protected files.
Start -> Help and Support -> search on "EFS".
If you don't know about EFS then it is likely that you haven't used EFS,
so for you it is a non-issue as yet.
3) If a few apps have to be re-installed after
the restore because they are tied to physical disk locations (games
often do this), I'm OK with that.
The drive letter assignments won't change. In DOS and Windows 9x/ME,
drive letters were assigned in the same order they were discovered by
the BIOS: each physical drive was scanned to note the primary and active
partition on that drive to assign a drive letter to it and then the next
physical drive got scanned, then the scanning returned to the first
drive and assigned drive letters to all of the logical drives defined in
an extended partition on that drive and then the next physical drive got
scanned for an extended partition. The motherboards IDE ports got
scanned first and then the BIOS was used for a SCSI card to find any of
its drives. Driver-supported drives, like CD-ROM and DVD-ROM drives,
got assigned drive letters last. A: and B: were reserved whether you
had floppy drives or not. Other than floppy drives A: and B:, Windows
2000 and XP don't base drive letter assignments on the BIOS scan order.
Part of MBR (master boot record) on each drive is a disk signature, a
unique alphanumeric string given to each drive (after it is initialized
within Windows). Windows can then identify the drive no matter if you
later move it around on the IDE/SCSI/SATA controllers or change which is
master and slave. That way the drive letter assignment follows the
physical drive. You might have C: on the motherboard's IDE-0 port as
master and then move it to the IDE-1 port as slave and it will still be
drive C:. It is possible the disk signature gets corrupted or
duplicates another, like when you move drives between hosts, so you need
to use the Disk Management applet to rescan the disks to ensure each has
a unique disk signature or gets assigned one (i.e., the disk signature
bytes in the MBR get updated). These disk signatures are also recorded
in the registry (and how Windows knows what drive letter to assign to
which physical drive).
If all you create is one primary partition on a drive then its drive
letter won't change. If you have multiple primary partitions, whichever
is the currently marked active one is the only one used (and how you can
get multiple operating systems easily installed on a host and switch
between them, although there are other more pollutive methods). If you
change the order of the primary partitions by moving them around using,
say, PartitionMagic, so the first primary partition becomes the second
primary partition, that still shouldn't hurt because only one of them
will be marked active and used at a time. I don't recall if you can
move logical drives around within an extended partition. I suppose it
is possible to move the partitions around so C: became D:. Then comes
along Windows XP's ability to define dynamic volumes using unallocated
space in a RAID-like Spanning volume to really confuse things. There
are utilities, like DriveMapper included with PartitionMagic, that will
search for references to the old drive letter and change them to the
desired drive letter (should drive letter assignments get jumbled), or
you can do that yourself by searching the registry and checking the
config files for your applications (very time consuming). Sometimes it
is just easier to save your data (if it is in the same install path as
the application), uninstall the application (if possible), and reinstall
it in the same spot to get it to then use the changed drive assignments.
But if you have 100 GB of programs to install, it might take less time
to edit the registry and config files than to [uninstall and] reisntall
all those applications. I could never figure by Gates decided to use
drive letters (Unix doesn't have them) but then this stupidity is a
carry-over from way back when he conned IBM into using his modified
Seattle DOS.
4) Having logical and physical backups is certainly suspenders and
belt - but, given my comments above, it sounds like the logical
backup is good enough. What I'm trying to do is have complete
enough backups that can be totally automated, without paying lots of
bucks per machine for a backup program, because eventually this will
go on 3 machines. I'll pay the bucks if I have to, but I haven't had
to yet, so I'm testing whether XP really requires high-power software
just to be able to recover from a disk crash. Not sure whether
you're saying it does or doesn't.
Logical backups are best for data backups. There's not much point in
backing up applications since you can simply reinstall them again (and
apply updates thereafter to get to the same version as before). That's
why when you define the folder and files to include in your logical
backup that you needn't bother to include the applications or even the
OS itself. I would, however, select to save the System State as this
includes the registry files (I don't remember if you have to be logged
in under an admin account to get the registry files for all accounts but
it sounds a plausible security mechanism). The NT Backup included in
Windows 2000/XP is, er, was a crippled version of Veritas' Backup Exec
Desktop. Veritas has since sold off that product to Stomp Inc where it
is now called Backup MyPC. I still have the old v4.6 of Backup Exec
Desktop since the sold-off version seems to have been mostly updated
just to replace the copyright and company info within the program's
files (i.e., no bang for the buck). Besides file and tape support for
backup filesets, I can also use removable media, like CD-R[W].
My father runs a business from home as a mechanical contractor and all
he does is logical backups. That is sufficient for him. When I setup
his machine initially, I created disk images. Whenever I do major
changes on his system, I save disk images. This allows for very quick
disaster recovery. I use CD-R media instead of using another drive
simply because I do not want to rely on a mechanical device to provide
disaster recovery. Hard drives go bad (i.e., mechanicals) and can be
damaged (from handling/shock or surges frying their interface
circuitry). With CD-Rs (eventually to be supplanted by DVD-R), I don't
have to worry about a bad drive as we can just get another one to
continue using the backup media. The logical backups for data (which
use tape so restores are slow but you can get a lot on tape and restores
shouldn't be something often needed) are sufficient to recover the data
from end-user mistakes or hardware failure. But recovering a full
system disaster (fire, hard drive crash, theft, etc.) by having to
reinstall the OS, all applications, and then walk through all the full
and weekly and daily incremental backups will take a L-O-N-G time during
which you cannot do your business (i.e., expensive!). That's why I like
disk images for quick recovery back to some snapshot you took of the
system. Instead of taking days to recover, you recover in a few hours
(depends on how big an image you are restoring). Downtime is very
expensive in business. Your customer really doesn't care about your
woes when they want your your job quote. The disk image is a snapshot
so you will still have to perform some logical restores for data that
was changed since then but that goes a lot quicker than having to go
through all the full, weekly incremental, and daily incremental backups
you have done since the beginning of doing backups.
Logical backups are good for restoring a few files. If the files are
all recently changed and backed up then even restoring several thousand
won't take that long, either. But having to restore a complete set of
data files through all tapes can be very expensive. It depends on what
backup media you use. Tape is slow but capacity is large (compared to
other traditional removable media). CD-R or DVD-R is nice, doesn't rely
on drive mechanicals for usability of the backup media itself, but won't
have anywhere near the storage capacity of a hard drive. Hard drives
are very quick for backups but you are risking loss due to mechanical or
electrical failure of the device, so use them for only as far back for
your incrementals as you are willing to lose data, like for your daily
incremental backups (or perhaps even for your monthly incrementals if
you are willing to lose data back that far).
And ALWAYS include the option in the backup program to VERIFY your
backup after it got created. This doubles the time to perform the
logical backup, but what the hell good is a backup that you find later
is unreadable? I've seen people take the easy route of not enabling the
verify option because they whine it takes too long, until later when
they try to perform the backup and find the tape, CD-R, or drive won't
read a portion of its media and that highly critical file is now
completely lost (or the required latest version is lost).
5) By all means, be verbose. I'm glad you have the luxury of doing a
clean install. You either have way fewer apps, or way more time 8-}.
Seriously, for the 'big" machine (the one my wife's tax prep business
runs on, with over 10 years of old apps that have to work, no chance
of porting them and moving the data - it would take days and you
still wouldn't be sure it's right), I'm contemplating a fresh install
into a different partition, and dual-boot. The current apps get
re-installed, the older apps maybe just work (financial apps are
light on their O/S use so they might work), or she'll have to reboot
under 98 to get them to work (assuming, and this is my fear, that the
old O/S doesn't suffer SW rot over time). On 'my' machine (also
win98se, but newer, smaller number of apps, and very 'clean'), I'm
contemplating an upgrade as I think if it will work anywhere, this is
a prime candidate, and after all, it is faster. However, any advice
you have would be appreciated. Note that I'm quite IT literate, built
all my machines myself (HW and O/S install), and keep them very clean
(no rogue or adware, etc). They're actually quite stable, and my
primarly motivation for upgrading is that newer SW isn't really 98
-friendly anymore.
Then I would say go with the upgrade, er, migrate from Windows 9x to
Windows XP. Even if you run into a problem later, it sounds like you
have the wherewithall to correct the problem, like reinstalling just
that application, changing its configuration in files or the registry,
or installing the appropriate motherboard, video, and sound drivers.
There are already so many problems when using Windows and various
applications on it that I'd rather spend the time up front than do it
later because I'll already be taking care of other problems later (even
if it were a fresh install). Consider it preventative maintenance, like
changing the oil in your car rather than waiting for a disaster or a
dental checkup instead of waiting for the pain to make you go in.
Sometimes you don't save any time when executing preventative measures
but they are often easier to implement than the effort involved in
recovering from a disaster. Replacing your laundry washer hoses every 5
years is easier than cleaning up your basement because the hose burst
and the expense of finding out that flooding wasn't covered by your
insurance.
It all depends on how much time you have and how secure you feel in
doing a migration rather than a fresh start. If you feel you have your
butt covered with backups and have a stable system and have prepped it
for the OS migrate and checked all your hardware and software is
compatible (or claims it is) then go for it. However, disk imaging
software really isn't that expensive to provide for very quick recovery
(back to what was usable before). The other choice is to mirror your
drive(s) onto other drives (that are NOT otherwise enabled in the
system). RAID-1 for disk mirroring only provides for hardware recovery
in case the primary hard drive fails. It does NOT provide a backup
function to get you back to where you were before since, obviously, the
mirroring was also occuring when you did the OS migrate. Mirroring is
for hardware disaster recovery, same for RAID-3 and -5, and not for
*BACK*up (as in going back in time). However, you can use disk cloning
software to mirror your production hard drives onto backup hard drives
and then disconnect the backup hard drives or remove them. Remember
that this procedure will use as much space on the backup hard drive as
for the source hard drive. You are cloning the drive. Disk images will
compress the data so its fileset will occupy less space; for example,
your 120 GB hard drive would occupy 120GB on a backup hard drive when
cloned, but maybe only 60 GB are actually inuse on that hard drive so
you could possibly use a program that only clones 60 GB onto the backup
hard drive and yet a disk image which incorporates compression (and
skipping of unused sectors) might only occupy 20 GB on the backup hard
drive, so using disk image software could let you put 6 snapshots on
that backup hard drive instead of just 1 or 2. My 63 GB sized C:
partition has 36 GB used on it and the disk image fileset is only 9 GB
big. That's a pretty big space saving. With a cloned drive, you can
swap it in and immediately be back up to speed. With a disk image, you
get the space savings but it does take longer to restore (it took 1 hour
to create the disk image fileset of 9 GB for the 36 GB used in my 63 GB
C: partition, but the time depends on your hardware's performance).
Since you are using NT Backup and presumably using the ASR wizard in it
to create a full backup for disaster recovery, and since you are backing
up [temporarily] to another hard drive, then disaster recovery (to
restore back to Windows 98) shouldn't be a real big pain. The speed of
the hard drive will help (but I wouldn't use it as the backup media to
archive a permanent or long-lasting snapshot). And be damn sure the
option to VERIFY your backup is enabled (I don't recall if the ASR
wizard asks you or you enable it in the options beforehand or can change
it during the ASR process). Disconnect that backup hard drive during
the OS migrate, and keep it safe from the kids, dogs, visitors, or
anyone else so it doesn't get shocked physically or electrically.
As a caution, yank the Cat-5 cable from your NIC if you are using DSL or
broadband cable for Internet access. Pull the phone cord if you are
using a modem for dial-up. It can take just 20 minutes of being online
to get your host hijacked or infected before you have setup adequate
protection. If Service Pack 2 is slipstreamed into your Windows XP
installation media then the included firewall will be enabled by
default; otherwise, you are susceptible until you get the firewall up or
installed and active and get your anti-virus software installed and
updated. Disconnect, do the OS migrate, ignore or cancel any activation
or updates suggested, get your shields up, and then connect to do the
activation and updates.