TI of Data Upload

[dharav]

Hello All

I faced strange behavior where in one server TI takes 13 seconds to finish the process and same TI under same environment but in different server it is taking 800 seconds. When process completes it runs sucessfully.

I re check script and also paste it same script from other server.

I restart the server also but there is same behavior. Could you please throw your input?

Thank You

Dharav

[jim wood]

CPU's..... Memory....... Users.... There could be many reasons. You need to give us more information that.

[dharav]

Jim

Could it be the reason?

[img]

M.JPG (28 KiB) Viewed 6257 times

[/img]

Thank You

Dharav Patel

[Wim Gielis]

Same version of TM1 on both? Same entries in the TM1s.cfg file?

What happens in the TI process, for example, Data Source is a flat file on a location that one server cannot easily go to?

Did you time Prolog / Metadata / Data / Epilog, breaking down total run time of the process?

If you reduce the size of the data source, do you see the same kind of effect or not ?

And so on. Try to break down the problem in smaller bits to see what's going on.

[blackhawk]

A cause in disparity may actually be the feeders re-processing as well. If nothing was changed on one server, the feeders don't need to be recompiled and you get the faster response, but on the other server that may not have been the case. What is the amount of time it takes for the feeders to process for the system?

Also, what is the configuration of the PersistentFeeder setting in the tm1s.cfg file? Are they both the same?

[dharav]

First of All , Thank You all for valuable inputs.

1) I turn off the PersistantFeeders in cfg file to F

2) Re write the feeder. It was overfeeding so it was consuming memory and process was taking time to upload.
once i modified and re start, now it's taking 13 gb instead of 79 gb. Process runs in few seconds.


I want to know about the feeder concept in following case (as i find in this issue):

Suppose there Two Cubes A and B

Cube A:

dim1: MetroHierarchy (all metros of a region)
dim2: versions
dim3:Counters
dim4: Measure A

Cube B:


dim1: MetroHierarchy (all metros of a region)
dim2: versions
dim3:Counters
dim4: Measure B


Cube C:



dim1: MetroHierarchy (all metros of a region)
dim2: versions
dim3:Counters
dim4: Asset Type
dim5:PCHierarchy (all cities of a metro)
dim4: Measure B


In the above structure,


When i feed Cube C from Cube A,

[{'Jan',...,'Dec'}]=>DB('C',!D1,!D2,!D3,'ALL ASSET ACCOUNT',!dim1('ALL PC'- i want to feed all metros in pchierarchy so i replaced d1),'TOTAL');

Cube C shows the data and Feed perfectly for assets account which lies in both Cube A and Cube C.

> I want to know, how it feeds the data to cube C for the asset account of Cube B without writing feeders inside cube B for cube C? (I just hit with this above one feeder and it works )<

Now, as i am not needed to save feeder in cube B for Cube C, It also improves performance at great extent.


Thank You

Dharav Patel

[Wim Gielis]

Is Asset account an element in the Measure B dimension ? Or input in the cube cells in 'B' ?
You need to be more detailed about what you want to achieve.

Also, please start a new topic. 1 topic = 1 question, on message boards like this.

[dharav]

@Wim : Thank You so Much for your time


=>I achieved it.

It was causing problem because cube A has Region LEVEL Hierarchy AND cube B has Metro Level hierarchy with Regions.
In the feeder, I was feeding Total Element of Metro Hierarchy due to which overfeeding appears. I feed Total Element of Region Hierarchy instead of Metro Hierarchy.

=> For the single cube feeder was feeding 2 seprate cube data. I would come up with all detail with new topic soon. I want to know the concept behind it.
It is working as good as we expect but before implemention i would like to know and learn about this concept.


Thank You

[Wim Gielis]

Good job !