Page 1 of 1

RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Fri Mar 23, 2018 2:36 am
by PeteB
Hi Everyone, (Sorry about not posting in the correct area previously.)

https://www.ibm.com/developerworks/rfe/ ... _ID=116669 and https://www.ibm.com/developerworks/rfe/ ... _ID=118320 are an RFEs for a TM1 commend to internalise the copy of rule values to values. The specific details are include in the RFE, and IBM have said it should be easy to do.

Today this RFE is the number 1 Planning Analytics RFE which is not slated for a future release, however, IBM have indicated they intend to reject the RFE as the capability already exists. I am hoping as many votes as possible will help IBM reconsider rejecting this RFE .

We typically use copy rules values to values at milestones through our planning process to keep a history snapshot of the plan to compare against updates. WE use bedrock to export the rule vales to file and then reimport. We use Hustle and RunTI to improve process times by over 50% to 1/2 an hour, but this is not acceptable performance. Here is a recent list of threads related to this issue:
http://www.tm1forum.com/viewtopic.php?f ... 656#p68656 Improve TI Processing Time
http://www.tm1forum.com/viewtopic.php?f=3&t=13901 Parallel run vs. ICX
http://www.tm1forum.com/viewtopic.php?f=3&t=13196 Copy version performance
http://www.tm1forum.com/viewtopic.php?f=3&t=13861 TM1RunTI and number of cores/threads

A TM1 in memory solution will be significantly quicker than our current approach of writing to disk and reimporting or the alternative, writing to a replica cube and back to original cube, and require less ongoing maintenance.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Sun Mar 25, 2018 3:30 pm
by Wim Gielis
I voted for it.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Mon Mar 26, 2018 1:06 pm
by lotsaram
Although this would be a great feature I'm certainly also sympathetic to IBM's position as to put it bluntly making data copy much easier will attract more fools but making such a process foolproof will be impossible. For example what should happen if the destination version is either wholly or partially rule calculated? Should the process error? Fail silently? Refuse to do any work at all? Complete successfully but only copy to open cells?
Or what about a multi-dimensional from/to mapping e.g 2017 ACT to 2018 BUD?

And 1000 variations thereof. I don't think these are trivial problems to solve. Especially if building in new functionality to the engine itself where it really needs to be bombproof. I can understand why IBM might not be so keen to do it.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Thu Mar 29, 2018 4:08 am
by PeteB
Hi Lotsaram,

Thank you for your input.
lotsaram wrote:
Mon Mar 26, 2018 1:06 pm
Or what about a multi-dimensional from/to mapping e.g 2017 ACT to 2018 BUD?
This is covered by the current RFE -
Allow developer to specify an optionally collection of, source and target tuples defining the 'slice' to be copied adhering to the following rules, keeping in mind that source and target cube can be the same but don't need to be:
- dimensions not mentioned in the source tuple are presumed to exist in the target tuple and any existing data for all elements of that dimension is being copied.
- dimension mentioned in the source tuple must either not exist in the target cube or be specified in the target tuple and the target tuple can contain elements of dimensions that don't exist in the source, implying that the dimensionality of the target cube needs to be equal to the number of unreferenced dimensions in the source plus the number of dimensions in the target tuple.
lotsaram wrote:
Mon Mar 26, 2018 1:06 pm
For example what should happen if the destination version is either wholly or partially rule calculated? Should the process error? Fail silently? Refuse to do any work at all? Complete successfully but only copy to open cells?
I will add your questions as a new comment to the current RFE. Potentially, error handling could be handled as a parameter for the function giving the developer the chance to control the error handling.
lotsaram wrote:
Mon Mar 26, 2018 1:06 pm
put it bluntly making data copy much easier will attract more fools but making such a process foolproof will be impossible.
Yes, this may be so, however the requirement to copy data will always exist, and TI developers will continue to make mistakes regardless of the options. Hopefully, the current threads on this forum related to data copying may become redundant and will be replaced by developers making mistakes using the new function.
lotsaram wrote:
Mon Mar 26, 2018 1:06 pm
I'm certainly also sympathetic to IBM's position
I am told this is purely a priority issue not a technical issue. The IBM comment has not come from the technical team. - ie the more votes the more likely the Product team will reconsider.

Cheers,

PeteB.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Thu Mar 29, 2018 3:58 pm
by lotsaram
Hey Pete,
quid pro quo I have voted for your RFE. My favorite RFE is this one for temporary objects to be accessible by other TI processes that are part of the same transaction as the TI that creates them and not exclusively by the creating TI.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Sat Mar 31, 2018 3:10 am
by PeteB
Hi Lotsram,

Voted for your favourite.

I had not seen to the "Cognos TM1" section previously I had looked for TM1 and found Planning Analytics, so I have now raised a second RFE for this function. I am not sure why IBM has the three different areas and if it makes sense to duplicate the RFEs.

Cheers,

Pete.

Re: RFE 116669 - TM1 Data Snapshot functionality - Copy rule to values in a TM1 cube

Posted: Fri Apr 27, 2018 3:18 am
by PeteB
Hi Everyone,

I have updated this thread with a link to the second RFE in the TM1 Cognos section https://www.ibm.com/developerworks/rfe/ ... _ID=118320

Regards,

Pete.