MV vs. Analytic workspace

Question

Good morning,
I read the document http://www.oracle.com/technology/products/bi/db/11g/pdf/comparision_aw_mv_11g_twp.pdf
and I not understand clearly the difference, this probably made due to a poor knowledge of English.
Someone just enlighten me?
AW generates views for the cubes and dimensions, allowing a SQL query, the MV too.
The MV directly contain any aggregates but it is also possible with cubes (you can define a level of aggregation total).
My assertions are they correct?
So what are the major differences between the MV and analytic workspace Oracle?
Thank you once again in advance,
David

Answer 1

The data for an 11g cube is stored in an AW. This includes both leaf and aggregate data.
This data in a cube can be accessed directly by an automatically generated SQL View. (This is sometimes called an 'ET View'.)
Optionally, the system can also generate an MV on top of the cube. This MV should be thought of as pure metadata on top of the cube -- the MV itself does not contain any data. The closest analogy is the
CREATE MATERIALIZED VIEW ON PREBUILT TABLE
statement. The MV defined on top of the cube provides a number of things.
(1) You can access the data directly by running SELECT * FROM <cube mv>. This is possible, but gives you nothing that you could not get by selecting from the ET View instead.
(2) It ties the cube back to the source tables so that the system will know if the data in the cube (i.e. the AW) is FRESH or STALE.
(3) It allows you to optimize the build of the cube by using FAST MV refresh if you change only a few rows in your fact table.
(4) It allows the system to rewrite queries against the source tables to access the cube directly.