Page 1 of 1

Cube with 1 TeraByte of Data

Posted: Wed Aug 01, 2012 2:38 pm
by cardantim
Generally speaking can TM1 support a cube with 1 Terabyte of data? I've heard not but wanted some other opinions on this.

Re: Cube with 1 TeraByte of Data

Posted: Wed Aug 01, 2012 2:52 pm
by paulsimon
Hi

I have worked on cube with a couple of hundred GB of RAM, but not anything as big as a Terabyte

My question is how do you know that the cube will be a Terabyte? The equivalent TM1 cube for a set of relationally stored data is quite often only a tenth the size of the relational data.

There is no theoretical reason why it can't be done, although the biggest Windows Server I have seen was 500GB of RAM. The performance would depend on whether the cube was doing simple consolidation or whether it involved rules, the frequency of updates, how many concurrent users, etc.

When you get to that level of data, you should consider whether you are doing Operational or Management reporting. I would suggest that you must be doing a fair degree of Operational level reporting. Therefore you might want to consider reducing the level of detail in the cube and therefore the volume of data, and using drill thru to the underlying relational source.

Regards

Paul Simon

Re: Cube with 1 TeraByte of Data

Posted: Wed Aug 01, 2012 10:39 pm
by PlanningDev
This is not official but I talked to an IBM Labs person at last years IOD and they were telling me that they have tested TM1 cubes of 1 terabyte in size. Again not official and no info on the level of testing but the conversation we had did go into TM1 and the terabyte + range of RAM. Actually if I recall correctly they were more worried about windows than TM1.