BO Universe : OLAP versus Relational

Dear expert,
I am new to this. And my current understading is:
- For using SAP as datasource, you can only have OLAP universe, and you can't really 'JOIN' data from several BW datasource. we can merge but not really 'JOIN' like those with data from relational database datasource.
- So, is having Infoset is the only way to 'JOIN' data from different datasource in BW system? Or there is other technique to mimic how we join data (or tables) from Relational database in a universe?
Thanks a lot.

Hi Didier,
Thanks for the pointer. Looks like Data Federator will require some additional purchasing. Do you know what is the ball-park figure for a company with 500-1000 SAP BO users?
Hi Khaled, thank you for your reply. I am doing APD for some cases, but I wonder if we want to have duplication of the data in BW everywhere we want to have it combined with other datasource (e.g. sales data with finance data, then sales data with inventory, etc), we will have many duplicate data. While if we have many universes who can treat each datasource as relational source (but the universe it self is not keeping any data), this will be better? <-- i am still exploring if this is true :). Infoset have similar behaviour but not very good performance.
Didier, just if you have, could you point me to a material to read about Data Federator?
Thanks a lot guys.

Similar Messages

Migrating from OLAP to relational

Hi,
I am supposed to work on an assignment that needs to migrate the data model from OLAP to relational. Generally the industry is moving the other way, i.e. from Relational to OLAP.
In my current assignment I have to re-rengineer an existing reporting that is carried out of Oracle Sales Analyzer on top of Oracle Express Server.
In the modified environment we are supposed to use Oracle 10g Standard Edition licenses from the customer, which (as far as I know) do not have the OLAP option.
So we have to come up with a Datamart that has to be in relational (star schema) form.Please correct me if there is any other possible solution also.
I believe some of you here would have certainly got a solution to my problem, that is,1) how to efficiently migrate the OES data into relational model. We have to report on Oracle Discoverer 10g, and ETL would be Oracle Warehouse Builder.
2) Is there a way to reuse the metadata for ETL or reporting purposes. I have a vague idea about the whole exercise, in which, I'll manually dig out physical table level details from OES setup and then pull those attributes into the star schema kind of model tables. But maintaining hierarchies and so many levels will be another problem in relational structures.
Please suggest.
Thanks
Goga

Unfortunately, OWB cannot use an Express database as a source. However, there is some good news. If you are using Warehouse Builder 10gR2 there is an new Expert available from download from the OWB OTN website that uses all the old OSA metadata files to create a completely new project with source and target modules. The source modules are the original dimension and variable flat files.
Because the model that is generated is a logical model you can implement it as either a relational or multi-dimensional schema (assuming you have not tried to use any features specific to a multi-dimensional model such as value based hierarchies or non-additive measures).
Goto to the OWB Exchange:
http://www.oracle.com/technology/products/warehouse/htdocs/OWBexchange.html?cat=EXP&submit=Search
Search for "Experts"
and the list one listed is:
OSA Migration Expert: Uses the OSA Global metadata files to allow an initial migration example of an OSA system.
Obviously this is not a supported utility so you are on your own, but hopefully this will at least provide you with a good starting point for creating your relational schema.
Hope this helps
Keith

Drill-Trough from Disco OLAP to relational without beeing prompted for pwd?

Hello,
we are developing a showcase for Discoverer for OLAP on a laptop. We also would like to drill-throug from an OLAP-Report to a relational plus (or viewer) worksheet. We do this like Marc Rittman in his article: http://www.rittman.net/archives/001252.html
So when we do the drill-through, we get to the login-page and have to provide our password. Can somebody imagine a way to get directly to the worksheet without the login-page beeing shown?
For it is only a showcase, it should rather be fast then secure.
Thank you,
Tobias

Hi Tobias
You must have not chosen the Use Connection ID radio button when you created the link using the Specify Worksheet dialog box. When you did this you were asked for the username, database and EUL to connect to but you were not asked for the password. Thus, when you execute the drill you are prompted to supply it.
However, if you were using Discoverer connected to an Infrastructure and have created public connections you can use the id of a public connection and enter it into the Use Connection ID box. This time, because the passwords of public connections are not known to the end users, the link should execute.
I know this is not what you wanted to hear but it's the way Discoverer interface between OLAP and Relational works I'm afraid.
Hope this helps
Regards
Michael Armstrong-Smith
URL: http://learndiscoverer.com
Blog: http://learndiscoverer.blogspot.com

Universe OLAP Options

XI 3.1 with SAP Integration
I have created a universe based on a BW query. According to the documentation, there should be OLAP options under Tools>Options but the only tabs I see are General, Save and Print/PDF.
What I am looking for are options to generate technical names, replace level prefixes etc.
What am I not seeing here?

Ingo:
SP2 for the intergration kit, BO XI itself or both?
I'll check with our administrator. I thought we were at the current SP.
Would it be maybe a good idea for you to create a sticky note in this forum describing how to determine install levels based on the version numbers for the various server and client components?

Help with universe with multiple related tables/loops

Post Author: bradwist
CA Forum: Semantic Layer and Data Connectivity
Hi. I am relatively new to Business Objects and have been putting together some universes for our use. Most things I can accomplish well enough. However, I've run into an issue that I think fits within this topic. I've got several related tables and I'm looking for the best way to construct the universe to support our queries. The best way to describe them is to say that the database has a table of Person records, and a table of Relationships. One Person may be related to another Person in a number of ways (sibling, spouse, parent, etc). This can be pictured as Person Relationship ID -
< PersonID -
< RelatedPersonID or Person1 -< Relationship >- Person2 where Person1 and Person2 really both point to the same physical table. I believe these tend to lend themselves to Aliases. However, each Person table really has a number of subordinate records in tables (addresses, phone numbers, etc). Address1 >--- Person1 -< Relationship >- Person2 ---< Address2 While this can be done (and I've done so), it's difficult to explain to some analysts why this construct is needed in the universe in order to answer a question like: Give me all of Bob's Siblings, their ages, and their addresses. This is a simplification, as our structures actually include other types of Entities (Person, Business, Government, Organizations), all of which can be related to one another. Is there a way that BO Universes will support this structure? Thanks.

exp system/manager file=exp.dmp log=exp.log full=y at the source database.
imp system/manager file=exp.dmp log=imp_show.log full=y show=y - create the log file without importing the data
edit the imp_show.log and extract the statements that are needed to re-create users,roles,alter user and grants.
Pre-create the tablespaces using SQL*Plus in the target database.
execute the modified imp_show.log(All DDl's)
imp system/manager file=exp.dmp log=imp.log fromuser=A touser=B
Re-compile all the Invalid Objects.
HTH
-Anantha

SAP BW Universe OLAP @prompt LOV using object reference not working

I have tried to follow Didier MAZOUÉ's "[OLAP universes best practices|https://boc.sdn.sap.com/node/20081]" document to create calculated measure in a OLAP universe built on top of a BW cube directly. The objective is to get the revenue figure from the same period of the previous year. I wanted to be able to have a LOV to specify a period when the object is used in a webi report.
When I hard coded the LOV as a data set as displayed in the following example ( {'[CD2007009]','[CD2007010]'} ) The prompt worked as expected.
(PARALLELPERIOD(.[LEVEL01],1,[0FISCPER].@Prompt('Period?',A,{'[CD2007009]','[CD2007010]'},mono,free)),[Measures].[0G_AERLOS])</EXPRESSION>
When I build my expression using an object referece as ( 'TimeFiscal year period' ), I was given an error msg saying "invaid prompt definition". I verified my spellings and syntax and they were correct based on the examples provided by the best practice document.
(PARALLELPERIOD(.[LEVEL01],1,[0FISCPER].@Prompt('Period?',A,'TimeFiscal year period',mono,free)),[Measures].[0G_AERLOS])</EXPRESSION>
Is there anything obvious that I missed. Or there are some extra configuration I have to do on SAP BW or Universe side?

Hi Mr. MAZOUÉ,
After a few days of struggling with this particular prompt issue, I finally figured out what was wrong with it. The object reference in the @prompt function is not able to process object references in a subclass correctly.
For example:
@Prompt('Period?','A','TimeFiscal year periodL01 Fiscal year period',mono,free)
The 'Fiscal year period' is a subclass under the main class 'Time'. If you use this prompt function in a MDX expression, it parse correctly. But it generates a prompting failure when referencing the object with the embeded prompt function in a webi report.
However, if I move the 'L01 Fiscal year period' object underneath the 'Time' class and change the prompt as @Prompt('Period?','A','TimeL01 Fiscal year period',mono,free). I was able to get a list of values when I designed a webi report using the object.
This seems to me is a limitation from the current OLAP Universe implementation.
Looking forward to your feedbacks!
thanks

Date time characteristic converted in a string in Universe OLAP

Hi there,
We are facing to the following issue. Maybe one of you has been already facing to.
We have created a BO universe on top of BI queries (based on MDX) and we make WEBI reports.
We note when we update/refresh the OLAP structure, all characteristics (Number and date) are converted/considered as a character and string and we loose in that way the time information.
This is very ennoying when sorting on time dimensions as WEBI can only sort on Alphabetic order.
Indeed, we would like to display the last 13 month of a key figure as from the last month.
As the time characteristic is automatically converted by the Olap refresh structure in a string, we can't display the last 13 months in logical/natural/calendar order.
Ex: NOV 2007 | DEC 2007 | JAN 2008 | FEB 2008 ....DEC 2008
Any idea :
- why the 'Refresh structure' command automatically convert into a string?
- how to force him not to convert to string and keep the date information?
- order date as the logical/natural/calendar one?
Beforehand, thanks a lot

Hi Ingo,
I'm using 0CMonth or 0CALMONTH which are SAP characteristics.
Do you know how to sort data by date and not by string?
When we recuperate infoobejcts from BI queriies, all objects are string and all LOV are sorted by string which don't help us.
imagine we have JAN 1996, JAN 1997,... FEB 1996 etc etc.
A nightmare
Laurent

MDX code in XI 3.1 SP3 universes for time-related metrics

Hi all,
I have a multiprovider in BW 7.01 and I have installed BOE XI 3.1 SP3, integration kit and all.
I need to calculate a metric (Sales Price) in this manner:
after selecting a period (month/year), for a customer, in a sales centre (both customer and centre are optional conditions).
I need to calculate these different values for the given metric:
- value for the period and for the same period of previous year, plus difference between both values
- accumulated value from january until selected period, same for previous year and difference between both values
- rolling sum of 12 periods up to selected period, same for previous year and difference between both values
And all of them according to different dimensions, such as geography, age, type, etc...
For reasons of volume, abstraction capabilities, etc. we need to use a multiprovider.
In your expert opinion, what would be the best approach for this?
Would you edit the universe and create new objects using MDX functions, such as PERIODSTODATE, etc.? If so, have you ever tried it? could you suggest a syntax, taking into account we don't have the MDX wizard yet in 3.1 SP3?
Is there some other approach that you've used, such as using formulas in Web Intelligence?
Regards!
David
Edited by: David Perez on Jan 10, 2011 4:21 PM

I found a workaround for the issue by patching and compiling the Axis2 kernel library. You need to download it form Apache, install Maven 2.0.7, set the default timeout in .../client/Options.java to a value that suits your needs (for me: 20min), and compile the whole thing using mvn clean install.
But this is obviously not the solution one wants. So, is there anybody with a REAL answer to the problem?
Regards,
con

O-R XMLtype versus relational table storage with XMLType

Hi,
I am working on a application which centers around ingesting/storing, editing and publishing a huge XML file (spanning several days of a calendar).
I am confused as to whether I should choose to store the XML as O-R XMLType (by registering the schema etc.) or just as an XMLType in a table.
The requirements are:
1. Farly frequent edits both and leaf and on branches.
2. Merging of XML when a new one comes in.
3. XML itself is of the order of 10s of Gigs.
4. Publishing as XML.
5. Middle-tier in JBOSS.
6. Validation before ingesting (could be dont in the middle tier).
7. Applying edits to the XML after it is stored and just before it is published.
8. XSLT Tranformations to an incoming XML to convert it to my format to be stored. (could be done in the middle tier)
This on Oracle 11g.
Thanks in advance.
Regards,
Vishal

Best place to start would be with an Oracle document at
http://www.oracle.com/technetwork/database/features/xmldb/index.html
Look at "Oracle XML DB : Choosing the Best XMLType Storage Option for Your Use Case (PDF)"

Universal object versus multiple varaibles

Hello
im currently developing some data management system wich stores and gets varius datatypes both primitive and higher level objects.
therefor i implemented a class 'datatype' compareable with these fields:
final static int TYPE_INT=0;
final static int TYPE_DOUBLE=1;
final static int TYPE_BOOL=2;
final static int TYPE_STRING=3;
final static int LEN_INT=4;
final static int LEN_DOUBLE=8;
final static int LEN_BOOL=1;
final static int LEN_STRING=0;
final static int MAX_STR_LEN=255;
public int type,size; //these are public for faster access. they are vital to the performance
Object value;now if i want to compare eg an int with a double in compareTo(datatype op) i have to perform massice casting and wrapper-instancing like:
case TYPE_INT:
   switch(op.type){
//not sure if this linde is 100% correct, i just want to ilustrate the idea
    case TYPE_DOUBLE: return (int)((double)((Integer)(value).intValue())-((Double)value).doubleValue())
}so i figured it may be better to have multiple fields in the datatype object, one for each primitive datatype like:
int intVal;
double doubleVal;
String stringVal;
and only set one at a time.
anyone has an idea how memory-consuming this may be?
does defineing a primitive data without init. it does consume memory?
its highly possible to have more then a thousend datatpy objects at a time so this is important to me.

//these are public for faster
access. they are vital to the performance
Object value;Are you absolutely sure about the above? Offhand I'd have to say that's pretty unlikely.
so i figured it may be better to have multiple fields
in the datatype object, one for each primitive
datatype like:
int intVal;
double doubleVal;
String stringVal;
and only set one at a time.Don't take this too hard but it looks like you are programming "FecalMatter V0.1"
anyone has an idea how memory-consuming this may be?
does defineing a primitive data without init. it does
consume memory?yes.
its highly possible to have more then a thousend
datatpy objects at a time so this is important to me.And if each of these Objects took up 1K an thousand of them would be 1 whole megabyte. That's sure to crush your Commodore 64.

Integrating OLAP Universe and Non OLAP Universe

Hi Experts,
I am facing a problem in integrating an OLAP universe and a universe on two dimensional database. Let me explain the scenario:
I have one BW query which gives plants A, B and C sales details. I have one more database (SQL Server), which gives plants D and E sales details. The requirement is to show all the plants (A, B, C, D and E) sales in Xcelsius Dashboard. For this i followed the below approach:
1. To bring SAP plants data, i created universe on BW Query (OLAP Universe)
2. Created universe on SQL Server tables to bring Non-SAP plants data. (General Universe)
Now my doubt is, is there any approach or functionality to join or merge these two universes (OLAP and Non-OLAP). I think we can merge Non-OLAP universes. If this is not possible then what is the best approach.
Can anyone help me on this.
Thanks & Regards
Ramakrishna Kamurthy

Hi,
I n order to merge prompts in Web Intelligence query panel, they must be identoical: same question, same datatype.
But you have to pay attention to the fact that only one list of values will be used for the twio data sources.
So it the two data sources have different keys (or tecnnical names in SAP) then one query will work and the other not.
Again if you try to synchronize those two data providers in WebI you will hit the same issues: keys and caption might be different.
So I suggest to do some SQL transformation on the relational universe to match keys and captions in SAP BW data source.
I recommend this because SAP BW does not support custom calculations that return string or dates: only numeric values are supported.
So in that case if you are sure that keys and captions will be identical in the two sources you will be able to merge prompts and synchronize both sources.
Regards,
Didier

How to use parallel period function in olap universe?

Hi all
I want to create a ratio with the measure of the last year with the function parallel period in a universe olap. What is the syntaxis for do this ? Thanks

Hi Didier.
Thanks for reply
I have a olap universe and i create a new object with this sentence:
<EXPRESSION>ParallelPeriod([0CALMONTH].[LEVEL01],1,[0CALMONTH].CurrentMember,(@Select(Ratios\Salidas stk tot caj)))</EXPRESSION>
When run a query with Rich Client this error appears: (spanish)
Se produjo un error en la base de datos. El texto de error de la base de datos es: La consulta MDX WITH MEMBER [Measures].[D29E6BC1-5AC0-4245-86,DA,DB,41,D6,CF,B3,52] AS ' ParallelPeriod([0CALMONTH].[LEVEL01],1,[0CALMONTH].CurrentMember,([Measures].[4GPT8D1S01T3MR5EZ57C98LJC])) ' SELECT { [Measures].[D29E6BC1-5AC0-4245-86,DA,DB,41,D6,CF,B3,52] } ON COLUMNS , NON EMPTY CROSSJOIN( { [0PLANT].[0021] } , { [0CALMONTH].[201002] } ) ON ROWS FROM [ZIC_MP01/ZIC_Q01] no pudo ejecutarse con el error Error desconocido. (WIS 10901)
What is wrong ??
Thanks

No Dashboard Design Query on OLAP Universe BO 4.0

Hi all,
the goal would be to create a Dashboard Design Query on a OLAP MSS Analysis Services Universe (UNX).
These are steps we follow to create the Universe (UNX) with IDT on BO 4.0.
1.) Create the Project
2.) Create the OLAP connection on OLAP Cube of MSS Analysis Services
3.) Publish the OLAP connection
4.) Create a Business Layer on OLAP Connection
5.) Publish the Business Layer into the Repository
Using that universe ( UNX ) with Infoview we have no problem to create a WeBi document.
Using Dashboard Design Version 6.0.0.0 Build: 14,0,2,594 to create a query using that Universe ( OLAP UNX ), the system doesn't list that Universe. It seems that it doesn't exist!!!
Someone has solved this issue?
Thanks in advance.
Luca Fatigati.

Hi Nikhil,
I have the same setup as mentioned by Luca Fatigati in previous post.
Trying to create Dashboard Design Query on a OLAP MSS Analysis Services Universe (UNX). But I am not able to .
Also I tried creating Explore on top of my UNX whcih is created on MSS Analysis Services Cube , ( OLAP UNX) , I am not able to see my UNX in Explore also .
Can you please let me know if this is not supported here as well . ?

What's the Difference Between OLAP and OLTP?

HI,
What's the difference between OLAP and OLTP ? and which one is Best?
-Arun.M.D

Hi,
The big difference when designing for OLAP versus OLTP is rooted in the basics of how the tables are going to be used. I'll discuss OLTP versus OLAP in context to the design of dimensional data warehouses. However, keep in mind there are more architectural components that make up a mature, best practices data warehouse than just the dimensional data warehouse.
Corporate Information Factory, 2nd Edition by W. H. Inmon, Claudia Imhoff, Ryan Sousa
Building the Data Warehouse, 2nd Edition by W. H. Inmon
With OLTP, the tables are designed to facilitate fast inserting, updating and deleting rows of information with each logical unit of work. The database design is highly normalized. Usually and at least to 3NF. Each logical unit of work in an online application will have a relatively small scope with regard to the number of tables that are referenced and/or updated. Also the online application itself handles the majority of the work for joining data to facilitate the screen functions. This means the user doesn't have to worry about traversing across large data relationship paths. A heavy dose of lookup/reference tables and much focus on referential integrity between foreign keys. The physical design of the database needs to take into considerations the need for inserting rows when deciding on physical space settings. A good book for getting a solid base understanding of modeling for OLTP is The Data Modeling Handbook: A Best-Practice Approach to Building Quality Data Models by Michael C. Reingruber, William W. Gregory.
Example: Let's say we have a purchase oder management system. We need to be able to take orders for our customers, and we need to be able to sell many items on each order. We need to capture the store that sold the item, the customer that bought the item (and where we need to ship things and where to bill) and we need to make sure that we pull from the valid store_items to get the correct item number, description and price. Our OLTP data model will contain a CUSTOMER_MASTER, A CUSTOMER_ADDRESS_MASTER, A STORE_MASTER, AN ITEM_MASTER, AN ITEM_PRICE_MASTER, A PURCHASE_ORDER_MASTER AND A PURCHASE_ORDER_LINE_ITEM table. Then we might have a series of M:M relationships for example. An ITEM might have a different price for specific time periods for specific stores.
With OLAP, the tables are designed to facilitate easy access to information. Today's OLAP tools make the job of developing a query very easy. However, you still want to minimize the extensiveness of the relational model in an OLAP application. Users don't have the wills and means to learn how to work through a complex maze of table relationships. So you'll design your tables with a high degree of denormalization. The most prevalent design scheme for OLAP is the Star-Schema, popularized by Ralph Kimball. The star schema has a FACT table that contains the elements of data that are used arithmatically (counting, summing, averaging, etc.) The FACT Table is surrounded by lookup tables called Dimensions. Each Dimension table provides a reference to those things that you want to analyze by. A good book to understand how to design OLAP solutions is The Data Warehouse Toolkit: Practical Techniques for Building Dimensional Data Warehouses by Ralph Kimball.
Example: let's say we want to see some key measures about purchases. We want to know how many items and the sales amount that are purchased by what kind of customer across which stores. The FACT table will contain a column for Qty-purchased and Purchase Amount. The DIMENSION tables will include the ITEM_DESC (contains the item_id & Description), the CUSTOMER_TYPE, the STORE (Store_id & store name), and TIME (contains calendar information such as the date, the month_end_date, quarter_end_date, day_of_week, etc).
Database Fundamentals > Data Warehousing and Business Intelligence with Mike Lampa
Search Advice from more than 250 TechTarget Experts
Your question may have already been answered! Browse or search more than 25,000 question and answer pairs from more than 250 TechTarget industry experts.

Universe on long text

I need to display very long text on Webi reports. However, the technical limitation of a BW infoobject is 60 characters. So Bex query from which the OLAP universe is created can only display partial text. A possible workaround is to write the long text to a transparent table during ETL and build a relational universe on the long text table.
(1) Is it possible to join an OLAP universe and a relational universe?
(2) Is it possible to write a WEBI report on relational universe?
Thank you for your input.

here you go,
build a relational universe on the long text table.
>> It will take 254/255 Characters Are you fine.
(1) Is it possible to join an OLAP universe and a relational universe?
YEP
But not in Universe level(Linked Universes may help) , you can work on both by combining @ Report.
(2) Is it possible to write a WEBI report on relational universe?
YEP
Gracias...!!

BO Universe : OLAP versus Relational

Similar Messages

Maybe you are looking for