Page 1 of 1

Best Practice for renaming Cubes

Posted: Thu Aug 15, 2013 9:03 am
by JamiseBondi
Hi all,

We are rolling out a new instance but are wanting to accelerate the build of this new instance based on existing instances. My logic is to use a replica of an existing instance and then go about renaming the relevant objects but therein lies the question:

I'm assuming that it's not advisable to simply go about renaming objects in windows explorer?
I'm happy to leave existing dimensions, views, subsets and processes untouched but would like to rename the cubes (ignoring control cubes) to be in line with the new instance.
What advice do you guys have regarding the best method of renaming existing cubes bearing in mind some of them have cube security (I don't mind if these rules need to be copied out to a text doc and then re-added later... don't let this spoil any creative suggestions).
As an aside, I'm also looking at removing an existing dimension from all cubes and had thought of using Performance Modeler for this purpose (the only task that I'll be using this tool for incidentally) to retain the cube data rather than having to create new cubes from scratch and import data.

Any ideas?
thanks.

Re: Best Practice for renaming Cubes

Posted: Thu Aug 15, 2013 9:47 am
by David Usherwood
Renaming cubes (and then restarting the server :) ) works fine - you need to rename the .RUX file as well. DB() references in the rules will need adjusting. The new cube will need the same security rights as the old cube - you could take a snapshot of }CubeSecurity and paste or DBSS the content against the new cube. If you use a cell security cube it's probably best to recreate, but you could try renaming that too.
Performance Modeller - I'd be quite surprised adding or removing dimensions in cubes which weren't built with PM worked properly - and the rules would almost certainly break. To be brutal, PM has gone down so poorly with the TM1 community that I wonder whether anybody has even tried.

Re: Best Practice for renaming Cubes

Posted: Thu Aug 15, 2013 10:00 am
by declanr
I set up a simple bit of Perl script a while ago that would cycle through the whole data directory (websheets etc included) and swap any reference from an old cube name to a new one through all the core directory files.
It always seemed to do the trick without any bother, as long as you made the obvious steps of stopping the service first, running it and then starting again. Remembering to back up EVERYTHING first!

I don't actually have access to that script currently but it was a pretty simple thing to write, If I recall correctly I think i stole the basic code from something someone else had written to change dates in .SQL files.

Re: Best Practice for renaming Cubes

Posted: Thu Aug 15, 2013 12:11 pm
by Harvey
I'm in the late stages of developing a product that does exactly this. It supports renaming cubes and dimensions, and automatically updates all the associated objects. It should at least give you a good starting point -- it takes about 5 minutes to create en entirely new instance of a server with all the names changed.

I haven't started marketing it yet, but would be happy to give you early access in return for some feedback and testing. PM me if interested.