[ale] OT: NT/XP/2K/2K3 Disk Imaging

[Brian MacLeod]

It is not the case, should you elect to use a tool called Sysprep. This is a
MS tool to allow imaging of 2000/XP/2003 machines by changing the SIDs in
the OS, and, optionally (but highly recommended) to reload all PNP drivers
(which most hardware is now anyway). You can put drivers in a dicrectory,
tell Windows where to look for drivers (registry key), and then run the
Sysprep tool. From there, use whatever imaging software you like (including
the venerable dd), and put that on a different machine. When it boots
Windows the first time, it will ask a few questions (if you hadn't provided
satisfactory answers), and realign its configuration to better match your
hardware.

It mostly works, but there are a few gotchas with regards to signed drivers,
and changes in the SCSI/IDE interfaces. Unfortunately the majority of my job
responsibilities are in maintaining this stuff.

PLENTY of resourse available online, though some will contradict each other.

For starters: http://www.appdeploy.com and http://microsoft.com

bnm


On 9/21/05, Jeff Hubbs <hbbs at comcast.net> wrote:
>
> Back in my age of pre-enlightement, the rule of thumb was that disk
> images made from one NT (or subsequent derivative) machine could be
> installed on outwardly identical machines, but doing so was discouraged
> because NT made installation decisions based on the hardware it saw at
> install time and that small running changes (e.g., different rev level
> within a mobo chipset) between outwardly identical machines would be
> detected at install time and different binaries would go on disk, so as
> to result in a deviation between the OS installed on an image-created
> machine and the OS that would be there had it been installed on the
> exact same unit. As a result, you could wind up with a machine that's
> crash-happy and you'd never be able to figure out why.
>
> Does anyone know if this is still the case with current releases of
> Windows OSses?
>
> As an aside, supposedly, Linux wasn't *as* susceptable to this because
> you could make allowances at kernel-config-time and beyond that, the
> kernel and modules would decide what they saw when they were first
> invoked, particularly at boot time. In other words, if you compiled an
> everything-but-the-kitchen-sink kernel and booted to it and/or started
> the modules, they would sense rev-level issues and switch themselves
> appropriately (such as the CMD640 IDE bug).
>
> As other people have, I ask here off-topically because of the general
> sharp-cookie quotient.
>
> Jeff
> _______________________________________________
> Ale mailing list
> Ale at ale.org
> http://www.ale.org/mailman/listinfo/ale
>
-------------- next part --------------
An HTML attachment was scrubbed...