Lotsaram wrote: by constraints think things like contribution dimension, limitation of single hierarchy, unable to re-use the dimension in another application, the security model getting messed up, etc.
...the reason I always point existing customers to "Central" applications first - there is no "approval hierarchy" or replacement of security on the dimension it uses, no restriction on re-use of views, etc...
For all intents and purposes, it is exactly what you get when you point at a server instance, except you can limit the objects to a specific collection. Having used applications in 9.5.1->, a lot of people associate them with restrictive, managed planning applications a la Enterprise Planning. That changed with 10.1, and you should find the Central application type doesn't add much, if any pain, to offer users.
Alan Kirk wrote:Until the design team gets out of the headspace that graphics and icons and swiping and swooshing are more important than performance, I don't want these guys replacing anything.
In the interest of full disclosure, and because I love painting concentric white and red circles on my back, - I am one of "these guys" (not the design team exactly, but PM's are even more culpable for the experience)... but I have to agree on one point. Performance is one of the most important parts of any technology solution (I used to work in software support, so stability and maintenance are up there for me as well). It is the reason I advocate the distributed architecture that gives me local data - if I have to watch spinning hourglasses and progress bars, just because I'm on VPN instead of in the office (at midnight, when I'm even happier to be sat at my laptop), I start to resemble Yosemite Sam on a failed rabbit hunting excursion.
Insight is built for the distributed architecture (yes, it works in connected mode as well, but not with the same performance you get from distributed) - and in the first release of it, the distributed architecture doesn't support those central applications that are easiest to spin off from an existing TM1 model. I recognize that this creates a gulf between the kind of experience you want for existing models, and what you actually get, but my first priority was supporting massive numbers of users contributing to high participation planning exercises - with that well in hand now, it may be that we can improve other areas. If we didn't think of what performance and stability would look like when you've used rules on the attribute control cube referenced from the expression you've defined your subset with, well, those brilliant customizations will hopefully all make their way into a test regimen. Of course, when someone has a great comparison for performance like "this view comes back in perspectives in 2 seconds and in Insight in 30 seconds", I would love to hear about it to help further that.
... and don't worry - no one is replacing anything yet. I'll suggest anyone who proposes such a thing spend a few months hanging out in this forum to boost their ego...
