Hi Rita,
You're a few steps ahead of where I would start, as you have got your 17.1 populated conventional setup on one set of servers, and the 17.1 blank AA on another set of servers.
Whereas I probably would have split it out as above ("(1) Move environments. (2) Reconfigure environment. Under no circumstances try to combine those two steps into one"):
1) Swing box the 17.1 populated convention setup over to the 17.1 new environment, keeping it as a conventional setup.
2) Then once on the new hardware, switch it from conventional to AA.
So you've got a choice at this point.
A) Back out the new environment, and continue as above.
Not such a bad option. You'd just need to revert your new blank AA environment back to a conventional setup for a period.
or
B) Look at what you need to do to do to swing box AND convert a from conventional to AA.
This is possible, but you'll need to do your own sanity tests on this, as we don't document this particular scenario.
But you would cover these steps:
* Copy the database
* Apply patches/options
* Copy the customisation files
* Copy the Attachments to their new Repository, and anything else which is file based not included in prior steps
* Run configuration to convert to AA
* Add on additional servers as needed
But this needs more thinking about. Definitely possible, and some sites may be able to advise here what they think of doing this.
It's not a horrible option, per se. If you do your testing, and run it through on a trial environment, I'm sure that it's not as bad as I may be worried about.
And it's potentially a better option, once you understand the process and have a solid run sheet. Why? Because it means you skip some genuinely unnecessary steps in the whole process.
Why is it not my preferred choice? Simply because both the "Swing Box" and the "Convert Conventional to AA" are two separate, well understood, well tested operations. They've each been run countless times by sites, and the kinks are mostly worked out, or known about.
But if you mash the swing box and the conversion into one routine, the onus is much more heavily on you to make sure that you test thoroughly this process, and have a solid run sheet for when you do it live.
Oh, and backups are your friend. If you have the opportunity to do this on virtual machines and take snapshots, it will make your testing a lot easier. And reversion from a failure is a 60 second event, rather than a two day panic.
Anyone care to chip in with advice?
Has anyone done a similar change?
Thanks, Kyle_R.