TM1 Forum

Dear all,

Here's the situation that has been bugging me. Recently we are planning to deploy TM1 to a department to do forecasting. Below are the information they are inputting when they are doing the forecast in the manual system.

Material#
Customer
CustomerPart# (Same Material for same customer can have different Customer Part#)
ShipToParty
ApplicationCode (Same Material Code can have different Application Code, for different customer, forecast)
Salesman
SalesOffice
Manufacturing Country ((Same Material Code can have different Manufacturing Country)
Rank (Eg A, B, C, etc)
Currency
Local/Transfer Business

As you can see, there are already 11 information (or dimension) is required in the cube. Plus some other dimensions like time dimension, version, year, and measure, this cube will have 15 dimension!

I had created the above cube, and I tried to simulate existing report, by putting a few dimensions like material, customer, shiptoParty, ApplicationCode, and Salesman. This crashes my TM1 environment (I'm using the local server which comes with the client). This worries me.

Have I not done the design well ? Can anyone advise if you were to fulfill this requirement, how would you build the cube?

Sorry for the lengthy post & thank you for your time.

I don't think the number of dimensions are problems. As long as you can satisfy the business requirement and sufficient RAM on your server, it is not an issue at all.
If it is a RAM issue (out of memory bla bla bla), the log will tell you at the point you are crashing.
I am curious about what you have done to crash the server?

Hi Macsir,

I was just trying to create a view. I put a few dimensions like material, customer, shiptoParty, ApplicationCode, and Salesman at the left pane, and Time dimension on the right (or top). I select all elements in the dimensions on the left, and select current Year (12 mths). When I click on Refresh, the system will hang. After a while, my local server is dead. It gives me some error msg.
You suspect this is due to insufficient memory on my local ? I'm running 4Gb RAM.

Hi,
Did you turn on the zero suppression?

Yes, I did
It seem if I put too many dimensions on the left, with all elements, the system crashes. If the lesser dimensions on the left, then it will work just fine. But then, how will I be able to get a report/view of all the Material#,Customer,EndUser, Application Codes, Salesman, etc that I had forecast qty ?

How many elements in each left dimension you are using in the view? Coz you should be aware that the final row counts are based multiplication of all your left dimensions' elements with non-zero measures. If there is too many rows which consumed much RAM, the cube view will tell you "view is too large, bla bla bla". In that case, you need to think about your filters or you can directly export the view by TI process.

What did the log say when you crashed server?

The log shows "4360 ERROR 2013-07-04 03:41:27.648 TM1.Server TM1 Server Abort: System Out Of Memory."
From the cube explorer, i can see this cube is only 48KB. But the view I'm trying to create might be too huge. I put 4 dimensions on the left, and 1 dimension has about 2,000 elements. This could have killed the server.
Maybe TI is really the only way out..

my bad. Just solved this mystery. merely 4 dimensions with 2000 elements should not have crashed tm1. It is the rule that is killing it. I did not put the SKIPCHECK!!
I turn on the TM1log and I saw system keep busying with the rules. So I check on the rules. I was so careless, did not put the SKIPCHECK.
After I put SKIPCHECK, it work just fine.