Hello

I must admit, I am a bit new to OLAP, so if what I am asking is widely know
knowledge: I am sorry, my books do not seem to cover it... (-> do you know
one that is really good?)

I am wondering about the following: My data is in an SQL Server 2000, I use
Analysis Manager 2000 (both SP4) to query. The data I am working on consists
of mainly 2 tables, a master (containing repaired devices) and a child
(containing records which describe what was done to repair each device,
simple 1:n relation)

My 1st Question: I need some measures for the master table, say the number
of devices in a given time frame - no problem. I also need some measures
that live in the child table, say the sum of all parts used to repair all
the devices.
My cube is built around the master table, now I cannot address the child
table to take measures there.
What I did is: I added some computed virtual fields to the master table that
contain the aggregated data from the child table - I wonder if that is the
right way to do, or the only way, or?
Should I built the cube around the detail table? Do I use several cubes for
such issues?

2nd Question: The database does not really consist out of the 2 tables
outlined above - say I got another child linked to my master. For the
earlier example, each master has ONE or more child entries. For this OTHER
child table, a master can have ZERO to many childs. Now, what happens is
when I link this table to my master in cube design, an inner join is
produced, and master lines without childs are taken out of the lists... can
I do a left join or such in Analysis Manager? Is there another way around
this (by having another cube and linking them or ?)

Thanks for help on those task, I do really appreciate it!
Ralf