Rules Vs TI Process (Static Data)

[tgaluskapatterson]

I currently have a cube with 360,710,144 memory used for feeders and 452,989,952 memory used for input data. In an attempt to improve performance of the cube which opens in about 20.6-20.7 second i decided to take the rules currently employed and have the TI process calculate the rules and create at static data. The strange thing is when i simply comment out the rule and save, then unload the cube, and restart the service the cube still opens in 20.6-20.7 seconds. I checked the StatsByCube again and the memory used by feeders is now zero however I am receiving no increase in performance on the cube. Is there any benefit to taking this rule and creating in the TI process? Also does memory having any relation to performance cause with a reduction in memory used of 360,710,144 i would expect a little better load time of the cube. If anyone has any ideas let me know? I realize 20 seconds isnt terribly long for a cube to open but trying to optimize as much as i can. Thanks!

[declanr]

Look at the VMM and VMT settings.

[tgaluskapatterson]

Should have put this in my initial question but the VMM and VMT are blank so currently set 65,536. Are you thinking if I provided more memory for the Stargate Views of this cube it would improve performance? Is there any benefit to redesigning the TI process and removing the rule? Thanks for you response.

[Duncan P]

BLB files. Delete them.

[EDIT] Sorry. I kind of assumed that you might have made the same mistake I did and edited the RUX file direct. If there is a corresponding BLB file there this will have no effect until you delete them and restart. However reading your post more carefully there are a number of things that indicate that this probably isn't the case for you. Firstly the 0 fed cells. Second, that the TI to populate the cells ran successfully.

Was it the entire rule that you commented out, or was it just part? If it was the entire rule did you also comment out the SKIPCHECK and leave the rule attached? If so this will turn change it to use dense aggregation mode.

[tgaluskapatterson]

I tried both ways. Examples listed below. I don't know the total memory of the server so I am checking now. Don't want to crash it if i start changing the VMM settings.

Test #1
SKIPCHECK;
#['Percent'] = ['Amount'] \ ['Amount', AR_Aging:'Total AR by Aging'];
FEEDERS;
#['Amount'] => ['Percent'];

Test #2
#SKIPCHECK;
#['Percent'] = ['Amount'] \ ['Amount', AR_Aging:'Total AR by Aging'];
#FEEDERS;
#['Amount'] => ['Percent'];

I don't know if more information is better in this case but the cube dimensions are listed below. I have tried to reorder the dimensions but they are optimized.
Name Memory Used # Elements
AR_Accounts 20KB 16
AR_Aging 16KB 12
AR_Customer 139272KB 370093
AR_LoadVersion 32KB 46
AR_Measure 13KB 2
AR_Territory 61448KB 33359

Thanks for the advise.

[macsir]

1. Has your proccess successfully created static data in this cube?
2. Try to delete the rule and check.

[tgaluskapatterson]

I have created a process to populate the Amount and Percentage with Static Data. Unfortunately it does not appear to improve performance as the cube still retrieves data in about 20 seconds. I also took your suggestion and completely deleted the rule saved and restarted the service. Seems to have no impact. Thanks for the ideas though.

I also bumped VMM up to 10,000,000 kilobytes for this cube and restarted the service. Doesn't appear to have improved performance at all.

If anyone has ideas or articles on TM1 performance I would greatly appreciate it. Running out of ideas. Maybe this cube simply has to many elements intersecting.

[Duncan P]

You talk of opening the cube. However you can't open a cube, only a view on the cube. What are the characteristics of this view? Which dimensions (with how many elements and at what level) are on rows and on columns? Is suppression on for the view?

Also there are a few configuration options (in the TM1S.CFG) that affect the performance of view creation. What are your settings for those?

[tgaluskapatterson]

Sorry about that i am opening the default view.

Row Dimensions
AR_Accounts - 6 Zero Level, 3 Consolidated Level 1&0
AR_Customer - 1 Consolidated Level 3

Columns Dimensions
AR_Aging - 6 Zero Level, 2 Consolidated Level 2&1
AR_Measure - 2 Zero Level

Filter Dimensions
AR_LoadVersion - 1 Zero Level
AR_Territory - 1 Fifth Level

Suppress Zero is On.

I have attached the tm1s config file for reference.

Let me know if this helps or if you are looking for additional information. Thanks

[tgaluskapatterson]

Additional Info
AR_Territory - Memory Used 8192KB # Elements 33359 - Title
AR_Measure - Memory Used 2KB # Elements 2 - Column
AR_LoadVersion - Memory Used 4KB # Elements 46 - Title
AR_Accounts - Memory Used 2KB # Elements 9 - Row
AR_Aging - Memory Used 2KB # Elements 6 - Column
AR_Customer - Memory Used 2KB # Elements 1 - Row