1. What are the
differences between Personal, Shared and Secured connections?
·
A
Personal connection is created by one user and cannot be used by other users.
The connection details are stored in PDAC.LSI file.
·
A
shared connection can be used by other users through a shared server. The
connection details are stored in SDAC.LSI file in the Business Objects
installation folder. However one cannot set rights and securities on objects in
a shared connection. Neither can a Universe to exported to repository using a
shared connection
·
A
secured connection overcomes these limitations. Through it rights can be set on
objects and documents. Universes can be exported to the central repository only
through a secured connection. The connection parameters in this case are saved
in the CMS
2. What are custom
hierarchies? How can they be created?
·
Custom
Hierarchies are defined in a universe in order to facilitate custom drill down
between objects from same or different classes according to user requirement.
They can be created from Tools -> Hierarchies in the BO Designer.
3. What is a context
in universe? How are they created?
·
In an
universe, a context defines a particular join path between tables or a specific
group of joins for a particular query. Any objects created on a table column
which belong to specific contexts is naturally compatible with all other
objects from same contexts. When objects from two or more contexts are used,
separate SQL is generated and results are then merged in a micro cube. This
makes sure that no incorrect result is generated due to loop or any other join
path issue. Contexts may be created using detect contexts feature or manually.
They are generally created based on logical calculation and business
requirements, hence the detect context method is not very effective. To
manually create a context Go to Insert à Context, give the
context name and select the joins that should be present in the context. For a
universe contexts should be created in a way that all joins(except shortcut
joins) fall in at least onecontext
4. What is a chasm
trap? How can it be solved?
·
In a
dimensional schema based universe, we may have one dimension table joined with
two fact tables such that both of them are one-to-many joins(F >- D -<F
). In such a scenario, if we drag a measure each from both the fact tables
along with dimensions from dimension table, the value of the measures in the
fact tables are inflated. This condition is known as chasm trap.
·
A chasm
trap can be solved using 2 methods:
·
In the
universe SQL parameters, the option, generate multiple queries for each measure
needs to be selected. This will generate separate SQL statement for each
measure and give the correct results. However, this method would not work, if a
dimension (for example date) occurs multiple times in the result set due to
chasm trap
·
A
better approach is to put the two joins in two different contexts. This will
generate two synchronized queries, thus solving the problem.
5. What is a fan trap?
How can it be solved?
·
In a
universe structure, we may have 3 tables joined in such a way that, the 1st
table has a one to many join with the 2nd table, which in turn has a one to
many join with the 3rd table(A -< B -< C). In such a scenario, if a
measure is present in the 2nd table and it is dragged along with any dimension
from the 3rd table, the value of the measure will be inflated. Such a condition
is known as a fan trap.
·
A fan
trap is solved by creating an alias of the 2nd table and defining contexts such
that, the normal table is joined only with the first table, while the alias is
joined with both the 1st and the 3rd table. We would take 2nd table’s measure
only from the normal table and other dimensions of the 2nd table from the alias
table
6. What is aggregate
awareness? What is its advantage?
·
Aggregate
awareness function is used in scenarios where we have same fact tables in
different grains. Using this function we can define only one object for the
measures in the fact tables as
·
@aggregate_aware(highest_level,lower
level)
·
We also
need to define dimensions for associated granularities and define their
incompatibilities with the corresponding facts through the aggregate
navigation. This is accesses through Tools -> Aggregate Navigation
·
The
advantage is that in a Webi or Deski report when one drags the measure object
with the dimension object of a particular granularity, the measure column from
the Fact table of the corresponding granularity is selected in the BO default
Query. If we did not use aggregate awareness, we would need to define separate
objects for each of the fact tables which would be difficult to understand from
a user’s point of view.
7. What are the 2
different approaches of implementing aggregate awareness? Which one is better
in terms of performance?
The 2 approaches are as follows:
·
Aggregate
tables are built in the database, which contains the dimension fields(not
foreign keys) along with the aggregated measures. In the universe they are
present as standalone tables, i.e they are not joined with any dimensions.
Aggregate aware function is used to define both the dimensions and measures of
such tables
·
No
aggregate tables are built in the database level. They contain the normal fact
table at different granularities. In the universe, aggregate aware is used only
to define the measures and aggregate incompatibility is set accordingly
·
The
first approach is better in terms of performance, since for the higher levels
of aggregation, all the information is obtained for a single table. However, a
large scale implementation of this approach in a dimensional schema is
difficult. In most BI projects, the second approach is preferred
8. What is a derived table? What is its utility?
·
A
derived table is a table created in the universe using an SQL Query from
database level. The columns selected in the query become the columns of the
derived table. A derived table can be used for complex calculations, which are
difficult to achieve in report level. Such calculations are done in query level
itself.
·
Another
use of derived table can be to access tables from a different schema through a
dblink.
9. How is a derived
table different from a view? Which one is a preferred solution?
·
A
derived table is present only in the universe level, while a view is created in
data base level. Generally views are preferred since, in its case the onus of
calculation remains on the database and it does not load the BO server.
However, in cases where developers do not have access to database, derived
table is the only solution.
10. How can we access one derived table from
another?
·
We can
access one derived table from anoth er
using the function @derived_table.
·
The
syntax is @derived_table(Derived Table Name)
11. What is Index Awareness? How is it
implemented?
·
Index
awareness is a property of the universe, by means of which values in the filter
conditions of the queries/data providers built from the universe, are
substituted by their corresponding indexes or surrogate keys. Generally the
values in the filter condition come from a dimension table (like country etc)
and we require a join with the fact table to get this value.
·
However,
if index awareness is implemented, this join is eliminated and the query filter
takes the equivalent index value from the fact table itself.
·
To
implement index awareness, one needs to identify the dimension fields which are
to be used in query filter. In the Edit Properties of the object, we get a Keys
tab. In this tab, the source primary key of the table from which the object is
derived needs to be defined as primary key, and the database columns for all
foreign key relationships with the other tables also need to be defined here.
Once this is done for all required dimensions, the universe will become index
aware
·
An
extended prompt syntax is available since BO 3.1. It is as follows
@Prompt(’message’,'type’,[lov],
mono/multi,
free/constrained/primary_key,
persistent/not_persistent,
{'default value':'default key'})
·
If the
indexes for the dimension object is defined in the universe and we define the
prompt condition on the object with the clause ‘primary key’ in place of free
or constrained, then the filter condition will convert the prompt values
entered to their corresponding indexes and eliminate the join with the
dimension table
No comments:
Post a Comment