ok. I've been developing cubes for quite some time now, but this kindof
surprised me-- i like it-- but it surprises me.

my co-worker-- he's been doing this for many more years than I have-- made a
cube with like 5 or 6 physical dimensions and like 25 virtual dimensions..
so the dimensions only have one level; etc.

I have probably 8 or 9 dimensions altogether.

it seems to handle ok... i just don't understand.. it's like totally
different than that which I am doing

is it normal to have that many virtual dimensions?

how can i measure the overall performance / significance of this?

it is drastically different than what I have been doing.. i like it.. i just
don't like not having hierarchies for dimensions. if all of the dimensions
are flat, can we make virtual dimensions that are hierarchial?

its like-- where i have a worker dimension with a level for manager and a
level for employee-- he has an employee dimension and then a virtual
dimension for managers.

times 20.

isn't it more efficient for cube browsing to have this in a single
hierarchial dimension? I know that his method is more efficient for cube
storage..

ps - we don't really have performance problems either route-- we just have a
VERY busy OLAP server and we have to be as cheap as possible.

-aaron

there is some input in this:

http://www.microsoft.com/technet/pro.../ansvcspg.mspx

my approach is NOT make the virtual dimension hierachical - I use virtual
dimensions to be able to cross tab a physical hierachical dimension.

as it says in the paper no aggregations is stored in the cube when dimension
is virtual.

<aaron_kempf (AT) hotmail (DOT) com> wrote

so that would mean:

a) more compact cube
b) faster cube processing
c) slower query response times

is that correct?

how should i quantify this? run the same report under both structures and
eval performance differences?

i just wish that there was a tool for testing a whole bunch of MDX
statements; and timing it..

-aaron


"Michael Vardinghus" <michaelvardinghus (AT) hotmail (DOT) com> wrote

Let me clarify, no aggregations are stored in the cube when the dimension is
virtual *IF* you use the default aggregation usage (which is set to
top-level only). If you change the aggregation usage to some other value,
then soft (or flexible) aggregations are created for the virtual dmension
levels which are not rigid, i.e. top-level only. You will notice that when
you create a virtual dimension that the aggregation is by-default set to
top-level only; for a physical dimension, it is set to Standard.
--
Dave Wickert [MSFT]
dwickert (AT) online (DOT) microsoft.com
Program Manager
BI SystemsTeam
SQL BI Product Unit (Analysis Services)
--
This posting is provided "AS IS" with no warranties, and confers no rights.

"Michael Vardinghus" <michaelvardinghus (AT) hotmail (DOT) com> wrote

Thank you Dave.

I really appreciate your help.

Sorry if I have been grouchy recently; I got fired from MSN because I was
complaining about Hotmail crashing every day (like that is news to anyone);
and I had some relatively severe Access bugs that nobody is taking
seriously.

I love ADP more than anything else in the whole wide world. It just seems
SOOOO much easier than VB6 or VB.net-- I mean-- leaps and bounds. I think
that it is a waste of time when you guys come out with products; and then
you won't commit to fixing serious usability bugs (Access ADP sometimes
randomly executes sprocs and 'create proc' (from BOL) gives me grief EVERY
DAY.).

And I really wish that ADP could consume OLAP data.

-Aaron




"Dave Wickert [MSFT]" <dwickert (AT) online (DOT) microsoft.com> wrote