in-place upgrade of NT

  • Thread starter Thread starter Ziek
  • Start date Start date
Z

Ziek

Just a quick question:
Why do most people view the in-place upgrade as the most risky method of
migration to AD?

It seems like if your NT domain model is fairly simple, then why waste so
much time/money/effort on doing a parallel migration?

The roll-back is so easy for an in-place if you simply keep a spare BDC
offline during the process; if anything goes wrong, bring the BDC back
online and make it a PDC.. how simple is that?

I don't see why most people view this method as extremely risky!
 
Why do most people view the in-place upgrade as the most risky method of
migration to AD?

The risk is in application compatibillity. When migrating there's a lot
more scope for testing application compatibillity and, more importantly,
minimum disruption to users.

It seems like if your NT domain model is fairly simple, then why waste so much time/money/effort on doing a parallel migration?
Click to expand...

A valid point. But many people don't have simple setups ;-) In which case,
this becomes decidedly more risky.

The roll-back is so easy for an in-place if you simply keep a spare BDC offline during the process; if anything goes wrong, bring the BDC back online and make it a PDC.. how simple is that?
Click to expand...

Pretty simple.

So yes, in small/ simple environments this is probably a very good way of
doing it.
 
I've done a number of upgrades, admittedly of pretty simple
setups with few problems.
The risk is in application compatibillity. When migrating
there's a lot more scope for testing application
compatibillity and, more importantly, minimum disruption
to users.
Click to expand...
Just to put the other point of view, it is best to test
outside of any current systems, and although I realise that
this cannot always be done, application compatability
testing should not really be done on the target system.
Also, depending on how you do it, an in-place upgrade could
have less effect on users than a migration.

I'd love to be able to do a migration, so that I can see for
myself which works best in the type of upgrades I have done.

Certainly, if you have a brand-new 2000 server, and you
don't want to downgrade it to NT to do an in-place upgrde,
then a migration is the only option.

Cheers,

Cliff
 
If you do an in-place upgrade, how does application compatibility become an
issue? You would not be upgrading any current PDC or BDC, as a brand new
win2003 box would take over the PDC role and undergo the actual upgrade, so
why would any application break? You would end up with a 2003 AD-domain,
with NT-BDC's still part of the network. If any applications are runing on
the BDC's at this point, we would have to determie how to migrate that
application to a win2003 server so that the BDC could be decommissioned..

I don't see any disruption to users with this in-place upgrade, other than
no PDC available for the amount of time it takes to undergo the in-place
upgrade..


ptwilliams said:
migration to AD?

The risk is in application compatibillity. When migrating there's a lot
more scope for testing application compatibillity and, more importantly,
minimum disruption to users.
Click to expand...
so much time/money/effort on doing a parallel migration?
A valid point. But many people don't have simple setups ;-) In which case,
this becomes decidedly more risky.
Click to expand...
offline during the process; if anything goes wrong, bring the BDC back
online and make it a PDC.. how simple is that?
 
Ziek said:
If you do an in-place upgrade, how does application
compatibility become an issue? You would not be
upgrading any current PDC or BDC, as a brand new
win2003 box would take over the PDC role and undergo
the actual upgrade, so why would any application break?
Click to expand...
Ok. You would need to install Win NT on the Win2003 box if
you wanted *it* to be the first DC in the Domain. You would
need to install it first as an NT BDC, then promote it to
PDC, then upgrade it back to Win2003 Active Directory.

The *only* way you can *upgrade* to a Win2000 or Win2003
Active Directory domain from NT is to do an upgrade in-place
of the PDC. A Win2003 box *cannot* act as PDC in an NT domain.

If an application uses only member servers in the NT Domain
then it will probably NOT break with an in place upgrade, I
don't think. But it is always best to test.

Cheers,

Cliff
 
Thanks, yes that was my intention: to install NT on a brand new box, then
in-place upgrade it to win2003 AD.

This way I don't touch any of the current prod boxes, and no impact to user.
Roll-back is simple because if upgrade fails, I bring back the BDC which I
have been keeping offline, and promote it to PDC, so the environment is back
where it started from..
 
Back
Top