Performance Modeler - Is anybody actually using it?

[vichammer]

I've been searching through TM1forum and other sources to determine if anyone is actually using PM in a Production capacity? Is anyone else having a terrible time trying to utilize this tool even in a test/dev capacity?

I have a client where we've utilized Applications mainly because they needed workflow. Everything was built and maintained using Architect. The environment includes BI and was recently upgraded to 10.2 BI and 10.1.1 TM1. The client has no solid technical resources competent enough to learn Architect and I've been hoping that Performance Modeler would cut down on the learning curve. I've attempted various tasks such as importing data and creating links with little to no luck. The application seems to throw errors when attempting to do much more that simply add elements to a dimension or create a cube view. The Applications piece works nicely and the addition of email notification is a big enhancement but really that's about it. Coming from a BI background and using TM1 for a few years I've been enjoying the speed and relatively bug free development process. This version of PM feels a lot like the initial release of Cognos ReportNet...nothing works reliably and there is little to no support or documentation to troubleshoot. Surely, the addition of PM SHOULD be a positive. The ability to import data from a Framework Package or Report seems like a good idea but this too seems to be plagued with issues. Creating a one time import works...just don't plan to ever modify the process. Anyways, just looking for any insight??? Maybe that's a bad choice of wording as everything in BI is now called Insight!!!

[lotsaram]

I think you provided a good summary.

There may be a few instances of brand new greenfield sites using PM but I cannot imagine any established TM1 application "upgrading" to using PM. (For al of the reasons you mentioned and many more.)

[David Usherwood]

I do like to quote tomok's take on this topic
http://www.tm1forum.com/viewtopic.php?p=34178#p34186
My own view is that PM could be useful in simpler apps either greenfield or within a TM1 server but covering a new requirement - but I'm going to admit that for those who have the TM1 skills and experience it's just easier to stay away from it.

[bunchukokoy]

The application seems to throw errors when attempting to do much more that simply add elements to a dimension or create a cube view.

This version of PM feels a lot like the initial release of Cognos ReportNet...nothing works reliably and there is little to no support or documentation to troubleshoot.

Anyways, just looking for any insight??? Maybe that's a bad choice of wording as everything in BI is now called Insight!!!

I hope there'll be more comprehensive materials and support for this PM if this is really the future of TM1.

(Sighs.)

[sachin]

PM can help with some of the tasks, which earlier were not doable in a quick way.

Ex -

1. Adding a new dimension to cube: Previously the only way we could do it in Architect was to delete the cube and build a new one
2. Removing a dimension from cube: Again the only way we could do it in Architect was to delete the cube and build a new one
3. Find all the cubes using a dimension: Just click on a dimension and it shows all the cubes that use this dimension
4. find subsets used in a view: Clicking on a view will show the subsets
5. Creating an accumulation link: It builds all the intermediate objects required
6. Moving objects between servers is easier: If you want to move a cube or bunch of cubes from one instance to another, PM will identify all the dependent objects, creates a zip file, which then can be extracted in your target. You do not have to manually figure out the file objects associated with the move

When it came to rule generation, we continue to do it by hand as we felt the code generated is not optimal using PM.

Also, watch out for orphan files in your data directory. Everytime you create an accumulation link PM generates lot of files in data directory - such as the intermediate cube, its dims etc. When you delete most of them go away except files with this naming convention Upload_SOME_NUMBER.blb. Over a period of development we now have about 30 odd such files! This has been logged as a bug by IBM.

In my opinion, PM has its own advantages. So I wouldn't dismiss it entirely. Folks coming over from Cognos' EP suite, PM will be a familiar interface to work with. I only hope PM gets smarter with newer release.

[lotsaram]

PM can help with some of the tasks, which earlier were not doable in a quick way.

Ex -

1. Adding a new dimension to cube: Previously the only way we could do it in Architect was to delete the cube and build a new one
2. Removing a dimension from cube: Again the only way we could do it in Architect was to delete the cube and build a new one

....

I think this is marginal/debatable. Yes PM will do this automatically which I don't disagree is a good thing but behind the scenes what is still happening is a new cube is being created, all the data is processed in from the existing cube and then the old cube is destroyed. It works in a demo or demo model because there's practically no data volume but throw in any decent data volume and it will choke up so the use is really limited to very early days prototyping in which case rebuilding a cube to add or remove a dimension is no problem anyway. It also doesn't clean up any rules or other code that may need to be adjusted for the meta data changes, this still needs to be done manually.

Still I'm hopeful for the future but my feelign is auto-generated code is never going to be any good and I would really like to see an equal amount of work put into enhancing the rule and TI editors to work WITH PM but allow people who know what they are doing to write proper efficient code.

[jim wood]

It works in a demo or demo model

I think you've nailed PM on the head. It's aim is to help EP users migrate over. Going forward it may become more usable, but not yet.