No Tabs to Display

[jv_oz]

Hello all,

I have a model in development where there is a long list of profit centres (850 odd categories in my prototype list) that roll up through a number of levels.  This list is then used to create the elist from the top of the list down 2 levels (150 categories) and a dlist from the bottom up (750 Profit Centres plus their parents) which intersects the bottom two levels of the elist.  I then use Access Tables and Cut Down models to limit the sparcity this would create and thus get models small enough to use.  I have reviewer edit enabled and multi elist slices allowed.  This means that the users could open one level higher than the lowest elist node and then edit all nodes in that group.

There are 30 million cells per elist item in the total model and the largest cube is 8.5 million cells.  After applying cut down models at the elist node, I expect to limit this to about one third of this size, with about 250 profit centres included in one set of cut down models (set being one level higher than the elist node).

I am currently prototyping this model and am running into a possible memory problem when I have more than approximately 15 profit centres included (where the rest are set to No Data).  The message I get when I open the grid is "There are no tabs to display".  Cognos knowledge base suggests this is a memory issue, however I would have thought I'd be fine with a model of that size when cut down are applied.

I know this is kinda complex, however if anyone has seen this sort of message and knows of a way I can update some xml setting, or change a CAC setting, or whatever, I'd really appreciate your input!

Thanks heaps

John

[moocher]

Memory issues would be my first suggestion

After cut down you expect it to be 1/3 the size? - 10m cells? This is still pretty large (best practice is 0.5m - but can be more) and you are then trying to open up 15 at the same time?!?

Also, is your access table/cut-down definitely working? - Have you used the epmodelsizereader.exe to confirm this?

[jv_oz]

An update. It was a memory issue and nothing I could do could get around it, other than changing the model (surprise, surprise I hear some of you chuckle!!).

So the model change that actually worked was to decrease the depth of the elist and force more categories inside the model itself and then not use multi elist item edit.  This allowed the model to open fairly quickly, however we then ran into performance problems when attempting break back.

As we still had well over 2 million cells per elist item, I have now convinced the client to forget trying to plan at the profit centre level (5,000 profit centres with 3,000 staff doesn't quite make sense) and to start "planning at the level that matters", which may be profit centre, but which will probably be significantly higher.

Cheers,

John

[jv_oz]

Further explanation on the issue is contained in this extract from the Cognos KB:

How do you determine the number of cells being opened in a Multi-Node View?

The number of cells for a multi-node view is calculated by the resolved level of access for those nodes in the multi-node view.

The following is an example of the resolved level of access (and cells) for a multi-node view:
•   There are 3 nodes, 1 cube, and each node has access to one cell in the cube.
•   Dimensions A, B, and C make up the cube.
•   Each dimension has 3 items in it. For example, dimension A has items A1, A2, and A3.
•   Further, each node has access to a different item on each dimension.
•   Node1 has access to A1, B1, C1
•   Node2 has access to A2, B2, C2
•   Node3 has access to A3, B3, C3
One might think that this would mean one cell per node(1x1x1=1), and since there are 3 nodes, there are 3 cells in the multi-node view (1+1+1=3).

What actually happens:
The access for the multi-node view is resolved for each separate dimension and in this case the multi-node view needs the 3 items from each dimension.

That is, items A1, A2, A3 on dimension A; B1, B2, B3 on dimension B; and C1, C2, C3 on dimension C.

Furthermore, there is not a different copy of each dimension for each node used in the multi-node view. The multi-node view has one copy of each dimension that includes all the relevant items that are going to be needed by the union of all the nodes.

The calculation is actually 3(dimension A) x 3(dimension B) x 3(dimension C) x 4(3nodes+total for aggregate in multi-node view) = 108 cells

Therefore, the only option is to restructure your elist so there are less children per parent or do not open the MNV's that are causing the out of memory error.