TM1 Forum

I would be interested to hear about others experience with this. Is there significant performance degradation in the TM1 environment when reporting directly from a cube? Is it generally best practice to have a separate reporting cube for that reason?

I don't believe that BI would impact a cube directly, however with the way that TM1 multi threads having a seperate cube, for me would be a good plan.

There is no direct impact unless you aren't using parallel interaction and/or you have very long read queries coming from the BI generated queries.

Think of connecting TM1 to BI as simply adding more read users to your cube. The big thing I notice is that BI can generate some long running read queries that you would not ordinarily see from within TM1. As such you need to be aware of this and design appropriately.

In short, in high usage situations from BI it may be better to have a separate reporting cube in TM1.

Personally I always have a separate reporting cube, I've come across some performance issues when using complex Report Studio reports against TM1 cubes.

Speaking of this member naming needs to be updated see this vid
http://www.youtube.com/watch?v=lvdGJU_wmBE

Also be aware that modifying the internal dimension order on will modify the order the users will see on the bi side.

See here:

http://www.tm1forum.com/viewtopic.php?f=3&t=8613

Hi,
we use TM1 primarily for BI and it performs admirably. I guess it really depends how big your model and your box is

P

Thanks everyone for your comments. This has definitely been insightful!