Handling Transactional-Level Data - Expert Cube Development with SSAS Multidimensional Models - page 128

To make the measures that we added in this way visible to drillthrough operations, we
have to be careful. The drillthrough operation must be usable from the main measure
group, but the Drillthrough Columns list will contain only measures from the new
measure group. We can specify measures from measure groups different to the one
specified in the Action Target section with no problems; the only constraint is that all
the measures in a drillthrough action must belong to the same measure group.

Drillthrough recap
In the following table, we can see all the combinations of drillthrough techniques,
storage modes, and dimension relationships and their impact on cube processing
and queries. It is a shame that ROLAP storage always has very slow performance if
you keep the data in the relational database.

Technique Storage Dimension

relationship

Impact on
processing

Impact on querying

Transaction
Detail
Dimension

MOLAP Regular Execute a

SQL SELECT
DISTINCT on
drillthrough
attributes

Fast query.

Fact Fast query.

ROLAP Regular No operations in

cube process

Complex SQL query
to get the drillthrough
result. It also executes
a SELECT DISTINCT
query on drillthrough
attributes.

Fact No operations in

cube process

Complex SQL query
to get the drillthrough
result.

Alternative
Measure
Group (only
for numeric
attribute)

MOLAP N/A Executes SQL

SELECT queries
on the fact table,
without using
any SELECT
DISTINCT before

Fast query.

ROLAP N/A No operations in

cube process

Complete scan in the
fact table. Dangerous
with large fact tables.