Criteria for defining OLAP cubes

Hi,

I am considering create a physical OLA cube in my DW.

I hace OLAP analysis that need preagregate data to work faster.

I have read and heard that there are some question to take in account before to create a cibe.

1) Space and processing time. As a criteria if you have more than 10 dimensions, the cube can grow a lot and it takes long to update, is it true??
2) You can have a lot of values in some cells of the cube where the join of dimensions don't have data
3) It makes a longer ETL or actualizacion process: source files -> ETL -> relational DB of DW -> agreggate -> physical cube

More things to take in account?
Is there any alternative to cubes?

Thanks in advance