Solving the Mystery of Data

Provider Synchronization
Stephen Bickerton October 2002
Noscentia
Presentation Information (Hidden Slide)

Author: Stephen Bickerton

Company:Noscentia
Presentation title:
Solving the Mystery of Data Provider Synchronization
Presentation abstract:
Learn tips and techniques for solving the mystery of data provider
synchronization with Business Objects reports. This session will
focus on the resolution of common synchronization problems
experienced by everyday users of BusinessObjects. See a
demonstration of an advanced reporting technique that creates
complex local variables with extended syntax crossing multiple data
providers. Benefit from hearing tips and techniques for utilizing this
advanced reporting technique, based on real-life examples
provided.
Copyright © 2002 Business Objects - All Rights Reserved
2
Topics

What is synchronization?
Best practices
Tips and tricks
Common problems and solutions
Questions

Copyright © 2002 Business Objects - All Rights Reserved

3
What is Synchronization?

Synchronization
Report developers establish relationships between two or
more data providers (or micro cubes of data)
To combine and display in a table, cross-tab or chart
Common values will be displayed as one row
Uncommon values will also be displayed

Copyright © 2002 Business Objects - All Rights Reserved

4
What is Synchronization?
Query 1 from Universe Query 2 from Spreadsheet

Unique values
Copyright © 2002 Business Objects - All Rights Reserved
5
What is Synchronization?
Establish a Relationship between the Common Dimensions

Copyright © 2002 Business Objects - All Rights Reserved

6
What is Synchronization?
Establish a Relationship between the Common Dimensions

The common dimension

objects become synchronized
or linked in the “Define link
Between dimensions”
dialogue

Copyright © 2002 Business Objects - All Rights Reserved

7
What is Synchronization?
The Two Data Providers Are Now Merged

All values from both data

providers are displayed

Unique values
originating from the
‘historical’ data provider

Unique values originating

from ‘Q1 spreadsheet’
data provider

Copyright © 2002 Business Objects - All Rights Reserved

8
Topics

What is synchronization?
Best practices
Tips and tricks
Common problems and solutions
Questions

Copyright © 2002 Business Objects - All Rights Reserved

9
Best Practices

Synchronization is limited to dimensions only

Occurs at the micro cube level
Local variables of all qualifications, details and measures
cannot be synchronized

Copyright © 2002 Business Objects - All Rights Reserved

10
Best Practices

Linked dimensions should have the same data type

The data type of any object can be found in the data manager
Universe designers should adhere to certain design
standards in order that the data type of potentially linked
objects match

Copyright © 2002 Business Objects - All Rights Reserved

11
Best Practices

Measures will aggregate/roll-up (sum, count, average,

min, max) to the level of the linked dimensions

The object, Fed Tax,

is a measure

Copyright © 2002 Business Objects - All Rights Reserved

12
Best Practices

Measures will aggregate/roll-up (sum, count, average,

min, max) to the level of the linked dimensions

Copyright © 2002 Business Objects - All Rights Reserved

13
Best Practices

Unlinked dimensions from only one data provider can

be added to a combined block
Ideally, users should downgrade all unlinked dimensions and
details to details that are associated to the linked dimension

Detail objects should be

associated and
dependent on the
appropriate dimension
at the universe level

Copyright © 2002 Business Objects - All Rights Reserved

14
Best Practices

For stable synchronization, linked dimensions should

be at the lowest level of detail (like third normal form)

“Group Name” is linked but not at the lowest level of detail

So “Fed Tax” is incorrect when combined with “Payroll Quarter”

Copyright © 2002 Business Objects - All Rights Reserved

15
Best Practices

For stable synchronization, linked dimensions should

be at the lowest level of detail (like third normal form)

There is one “Group Name”

value for four values of
“Payroll Quarter”

One linked value to many

unlinked values can lead to
erroneous data

Copyright © 2002 Business Objects - All Rights Reserved

16
Summary of Best Practices

Synchronizing is limited to dimensions only

Linked dimensions should have the same data type
Measures will aggregate/roll-up (sum, count, average,
min, max) to the linked dimension
Unlinked dimensions from only one data provider can
be added to a block
For stable synchronization, linked dimensions should
be at the lowest level of detail (like third normal form)

Copyright © 2002 Business Objects - All Rights Reserved

17
Topics

What is synchronization?
Best practices
Tips and tricks
Removing null values
Prevent data providers from synchronizing
Common problems and solutions
Questions

Copyright © 2002 Business Objects - All Rights Reserved

18
Tips and Tricks
Removing Null Values

Synchronized dimension
values, unique to one data
provider, result in null or
empty values in the combined
block

Null values

Copyright © 2002 Business Objects - All Rights Reserved

19
Tips and Tricks
Removing Null Values, Accessing the Filter Dialogue
To create a defined
filter, select
Format/Filters from
the menu

Copyright © 2002 Business Objects - All Rights Reserved

20
Tips and Tricks
Removing Null Values, What To Filter…

“Department Name” is the

linked dimension and is at
the lowest level of detail

Copyright © 2002 Business Objects - All Rights Reserved

21
Tips and Tricks
Removing All Null Values - the Equality Join Effect

The filter logic should be

based on measures or
details, as they act in
association to the linked
dimension
Filter Logic
= Not IsNull(<Q1 Increase>)
And Not IsNull(<Number of
Employees>)

Copyright © 2002 Business Objects - All Rights Reserved

22
Tips and Tricks
Removing All Null Values - the Equality Join Effect

All null values have been

removed

Copyright © 2002 Business Objects - All Rights Reserved

23
Tips and Tricks
Removing Some, Not All, Null Values - the Outer Join Effect
=Not isNull(<Number of
Employees>)

Filter returns non-null values

of the measure object
“Number of Employees”
effectively removing the null
values generated from the
“Q1 spreadsheet” data
provider

Copyright © 2002 Business Objects - All Rights Reserved

24
Tips and Tricks
Removing Some, Not All, Null Values - the Outer Join Effect

Null values

Copyright © 2002 Business Objects - All Rights Reserved

25
Tips and Tricks
Preventing Data Providers From Synchronizing
BusinessObjects
will automatically
link all identical
dimension objects
when two or more
data providers
reference the same
universe

Copyright © 2002 Business Objects - All Rights Reserved

26
Tips and Tricks
Preventing Data Providers From Synchronizing

Report developers
may wish to
permanently unlink
dimension objects
across data
providers

Copyright © 2002 Business Objects - All Rights Reserved

27
Tips and Tricks
Preventing Data Providers From Synchronizing
Problem - unlinking dimension objects will prevent data
synchronization until the query panel is opened

To prevent the cubes

from auto-synchronizing,
the editable option should
be disabled for both data
providers

Copyright © 2002 Business Objects - All Rights Reserved

28
Topics

What is synchronization?
Best practices
Tips and tricks
Common problems and solutions
Synchronized dimensions not rolling up
Variables not compatible error
Incorrect aggregation
Defined filters with large row counts
Questions

Copyright © 2002 Business Objects - All Rights Reserved

29
Common Problems and Solutions
Synchronized Dimensions Not Rolling Up

Queries linked on ID / Employee ID and Department

/Group Name

Copyright © 2002 Business Objects - All Rights Reserved

30
Common Problems and Solutions
Synchronized Dimensions Not Rolling Up

Symptom - what
appears to be
identical data
displayed on
different rows

Copyright © 2002 Business Objects - All Rights Reserved

31
Common Problems and Solutions
Synchronized Dimensions Not Rolling Up

Employee ID
from data
provider one is
defined as data
type “character”

ID from data
provider two is
defined as data
type “numeric”

Copyright © 2002 Business Objects - All Rights Reserved

32
Common Problems and Solutions
Synchronized Dimensions Not Rolling Up - Solution

Solution - redefine the data type of the ID column in

the spreadsheet
Save spreadsheet as a .txt
Reimport the .txt into excel
Redefine ID column as text
Data type could also be
changed in the universe (not
advised)

Copyright © 2002 Business Objects - All Rights Reserved

33
Common Problems and Solutions
Synchronized Dimensions Not Rolling Up - Trailing Blanks

Symptom - linked objects with the same data type

not rolling up
State value “CA” appears over multiple lines

Copyright © 2002 Business Objects - All Rights Reserved

34
 Common Problems and Solutions
Synchronized Dimensions Not Rolling Up - Trailing Blanks

A query returns trailing blanks

Solution 1 - use a database function e.g., rtrim() (Oracle,
Sybase, Access 97), at the universe level or in free hand SQL
Solution 2 - select the Options command button within the
query panel to eliminate trailing blanks

Copyright © 2002 Business Objects - All Rights Reserved

35
 Common Problems and Solutions
‘Variables Not Compatible’ Message or Objects Grayed Out

BusinessObjects
prevents the
combination of
two unlinked
dimensions from
different data
providers

Copyright © 2002 Business Objects - All Rights Reserved

36
 Common Problems and Solutions
‘Variables Not Compatible’ Message or Objects Grayed Out
Report Manager with “Zip Variables dialogue with
Code” grayed out Insert command button
grayed out

Copyright © 2002 Business Objects - All Rights Reserved

37
 Common Problems and Solutions
‘Variables Not Compatible’ Message or Objects Grayed Out

Change qualification of unlinked dimension to detail

Associate the detail to the linked dimension

Copyright © 2002 Business Objects - All Rights Reserved

38
 Common Problems and Solutions
‘Variables Not Compatible’ Message or Objects Grayed Out

The detail should

be associated
with the linked
dimension at the
lowest level of
detail

Copyright © 2002 Business Objects - All Rights Reserved

39
Common Problems and Solutions
Incorrect Aggregation of Objects
The many to one relationship

There exists a many to

one relationship between
linked dimension
“Department Name” and
unlinked dimension
“Group Name”

Copyright © 2002 Business Objects - All Rights Reserved

40
Common Problems and Solutions
Incorrect Aggregation of Objects

The measure object “Gross Salary” does not roll up

correctly when displayed in a block with only the
unlinked dimension “Group Name”

Copyright © 2002 Business Objects - All Rights Reserved

41
Common Problems and Solutions
Incorrect Aggregation of Objects

Due to the many to one relationship, the MultiCube

function can be used to force the measure value to
roll up correctly

The MultiCube function is available from version 5.0.x

Copyright © 2002 Business Objects - All Rights Reserved

42
Common Problems and Solutions
Incorrect Aggregation of Objects

The one to the many relationship

There is a one to many
relationship between
the linked dimension
“Group Name” and the
unlinked dimension
“Payroll Year”

Each “Group Name”

value has many
“Payroll Year” values

Copyright © 2002 Business Objects - All Rights Reserved

43
Common Problems and Solutions
Incorrect Aggregation of Objects

“Fed_Tax” the measure, will display erroneous results

if displayed with “Payroll Year” the unlinked dimension
BusinessObjects cannot logically solve
this problem, as it only knows the
relationship of the Tax paid by each
Group Name
The linked dimension is not at the
lowest level of detail
MultiCube() cannot resolve the one to
many problem

Copyright © 2002 Business Objects - All Rights Reserved

44
 Common Problems and Solutions
Performance Problems - Data Providers with Large Row Counts

Synchronizing data providers with large row counts

can cause performance problems (calculation time)
This occurs when
Creating defined filters
Creating variables
Custom sorts
Ranking

Copyright © 2002 Business Objects - All Rights Reserved

45
 Common Problems and Solutions
Performance Problems: Data Providers with Large Row Counts…

Creating a defined filter with a large row count can

can cause the calculation time of the filter to exceed
the execution time of the query
Example:
Synchronizing 2 data providers with 220,000 and 50,000
rows, and 10 columns – calculation of defined filter over 30
minutes

Copyright © 2002 Business Objects - All Rights Reserved

46
 Common Problems and Solutions
Performance Problems: Data Providers with Large Row Counts

Work around
Create the defined filter when the data providers have been
purged or have a very small row count
Refresh the query and return a full data population
The defined filter calculation does not occur in the filter
dialogue and the calculation time is less than three 3 minutes

Copyright © 2002 Business Objects - All Rights Reserved

47
Summary of Tips, Problems & Solutions
Removing null values
Equality join effect
Outer join effect
Prevent data providers from synchronizing
Synchronized dimensions not rolling up
Data type issue
Trailing blanks
Variables not compatible error
Incorrect aggregation
Many linked dimension values to one unlinked dimension value
One linked dimension values to many unlinked dimension value
Defined filters with large row counts
Copyright © 2002 Business Objects - All Rights Reserved
48
Topics

What is synchronization?
Best practices
Tips and tricks
Common problems and solutions
Questions

Copyright © 2002 Business Objects - All Rights Reserved

49
Questions

Contact Information
Stephen Bickerton, Noscentia
(877) 722-7711,
www.noscentia.com
[email protected]

Noscenti
a
‘Knowledge through Information’

Copyright © 2002 Business Objects - All Rights Reserved

50