Seeing if a cube is unloaded.

[mincharug.shulft]

Hi ,
Can i know where you have seen whether the cube has been unloaded from the memory or not?
as its urgent for me.please help me gentle mans.

[Steve Rowe]

As far as I am aware it is not possible to see if a cube has been unloaded or not in the sense of looking at a flag.

If you have performance monitor running it may be possible to deduce that a cube has been unloaded by comparing stats at different times, doable in a TI but a challange to get the detection logic accurate I should think.

Cheers,

[mincharug.shulft]

HI,
Thanks for your reply.

When I opened the “}StatsByCube” cube to see the “Total Memory allocation” for cubes total
I am seeing the number like this ” 123787776”
So is it kilobytes or bytes?

Thanks.

[Alan Kirk]

HI,
Thanks for your reply.

When I opened the “}StatsByCube” cube to see the “Total Memory allocation” for cubes total
I am seeing the number like this ” 123787776”
So is it kilobytes or bytes?

The metric is actually "Total Memory Used", not "Total Memory allocation".

So let's see what the manual has to say about it:

The total amount of memory used by a cube, measured in bytes.

If that seems higher than you were expecting, then as I told you in the other thread you should looking for rules and/or feeders that may be holding the "unloaded" cube(s) in memory.

[rmackenzie]

When I opened the “}StatsByCube” cube to see the “Total Memory allocation” for cubes total

For a specifc cube, you can check 'Number of populated numeric cells' (in }StatsStatsByCube) for that particular cube (in }PerfCubes) for 'LATEST' (in }TimeIntervals) .

So what is going to happen next? Are you setting a flag that says whether the next user can log in and view their cubes... because you've made sure that because the last cube unloaded that there's now enough memory available to load some new ones?

Are you aware that your memory management strategy is seriously unusual?

There are some alternatives:

• archive off some old data to bring down the cube sizes
• rejig rules and feeders to only address certain slices of cube data so that your entire model is not being calculated unnecessarily?

[mincharug.shulft]

Thanks i got it.

[mincharug.shulft]

HI,
we implemented the TI script using cubeunload function and we are able to see the defference once we ON "performance monitor" BUT
why we cannot see the change of the RAM-Allocation in the Taskmanager's?

Could all gentle man's help me here ?.
thanks in advance to all.

Cheers.

[asutcliffe]

HI,
we implemented the TI script using cubeunload function and we are able to see the defference once we ON "performance monitor" BUT
why we cannot see the change of the RAM-Allocation in the Taskmanager's?

Could all gentle man's help me here ?.
thanks in advance to all.

Cheers.

It's because the TM1 process doesn't release the RAM back to the operating system. Instead it sticks it on it's own "garbage" pile for re-use. So as far as Windows is concerned, TM1 is still using that RAM.

Note, this question has almost certainly been answered before (probably in more detail than I can muster right now). A quick search always helps.

Also as others have suggested, it seems unlikely the approach you're taking is the best solution to your underlying problem.