Views in Dimensional Modelling
Hi
I need info on roles of views in dimensional modelling.,I have searched for this but no luck,,
Can anyone help me in this regard?
Are you looking to create views on Dimension model structure?
Yes. Dimension model structure are oracle tables. You can create view/synonym etc.
I hope this help.
Similar Messages
-
Hello,
I’m hoping someone can help me with understanding advantages and disadvantages if we want to build a universe on top of a fully normalized tables, compared to using a dimensional model with star schemas.
I’ve read some discussions here that say that it is possible to create a universe on top of normalized tables. Then, can we avoid building of dimensional tables (a data mart), and just use normalized tables? I would say that it is easier to use star schema dimensions and facts tables to build a universe, but our end users might ask “why do we have to go through building a dimensional data mart, if we can have same reports with hierarchies and drill-down functionality based on a universe built on top of our already existing normalized tables?”
Can you point me to some established best practices regarding using normalized tables to build a universe? Any documents with some examples for this?
Any expected difficulties during design/development phase of our universe, related to using normalized tables?
Any expected performance degradation if we use normalized tables compared to using dimensional tables?
Can I build universe more easily if I transform (modify) our normalized model (by using alias tables and views) to look like snowflake model?
I’m using BOE XI 3.1, tables are in Oracle 11.2.
Thank youFew Disadvantages that I usually face when building universe on Normalized tables,
1. Performance - Read operations have to suffer because indexing strategies do not go well with table joins
2. Derived Tables - Due to complex Queries/ Logic, most of the time, end-up creating derived tables, which doesn't use back-end table indexes and slows down the report runtime.
3. Normalized table/ Transaction tables may not always have proper cardinality and often results in Cartesian products
4. Normalized tables may not have tight referential integrity and may have to join more than one column and join on varchar, etc whereas, good Dimensional model datawarehouse will have proper keys/ integer joins and not always necessary to join on multiple fields
5. Often deal with Fan and Chasm Traps
6. Dealing with Database fields with nulls, blanks, date in numeric format, etc.,
7. No Facts, Dimensions separated and most of the time they are in same table
and More...
If performance is not a matter and building Datawarehouse is a big deal, then I will start building Universe on normalized tables by having the database diagram as reference for Joins, contexts, etc
Note: After dealing with universes based on normalized tables for few years (by creating views, complex sql queries data loading to tables and unv on these tables, derived tables), I ended up creating star-schema dimensional model (couple of months extra ETL work), users/ developers felt lot better when they have to create standard/ ad-hoc reports and they are super fast compared to previous universes. -
Dimensional Modeling on SAP Hana
Hello,
I am a newbie on SAP hana, so need help from experts / gurus.
Imagine the following Scenario ---> Oracle ERP is the source System and we want to build a Analytic Solution on Hana. Also assume that tables can be replicated into SAP Hana
Questions
1)For any kind of reporting / Analytics? Should we build a traditional Datawarehouse on Hana with all the features found in a star schema like surrogate keys, conformed facts and dimensions etc? or use only attribute views / analytical views provided as default and build datawarehouse model on top of this? In the second case, do we still need surrogate keys or how can we merge slowly changing dimensions?
2)Where can be push the transformation logic? Into ETL tool or use Hana provided sql script? and also where do we do this transformation, outiside hana where we do all transformations and load only data into normolized format? or within hana using scripts?
3) In any case is dimensional modeling really required on Hana? if not how can we map business process and do analysis on them? is it by means of only the views provided by Hana?
Your help is much appreciated.
Thanks!Hi Rajiv,
I'm no expert either and am just beginning to learn HANA, but here are my thoughts (community, please correct me if I am mistaken.):
1. HANA leverages existing data modeling theories, such as the entity relationship model. This model is simply a way of organizing tables in a way that makes sense for reporting. For example, the star schema, optimized for analytical reporting. The HANA view is similar to a universe, it allows the end user to connect to it from their BI tool. In the background, the modeler still needs to pull in tables, create joins, leverage primary and foreign keys, etc. So in short, traditional modeling practices and theories still come into play.
2. If extracting and transforming data using SAP's Data Services, the transformation of data will happen in HANA, using the Data Quality Management GUI.
Just as a side note, HANA has 3 methods of extracting
data:
Data Services
SAP Replication
Server
SAP Landscape
Transformation
3. Yes, dimensional modeling is still required. At the end of the day, HANA is still a database. We
haven’t changed the theory of data modeling, it still needs to happen, our approach is just different.
Hope that helps.
Kyle -
Designer - Dimensional modeling
Can Designer do Dimensional modeling ???
Welcome replies directly to my personal email.
Thanks!Hi Rajiv,
I'm no expert either and am just beginning to learn HANA, but here are my thoughts (community, please correct me if I am mistaken.):
1. HANA leverages existing data modeling theories, such as the entity relationship model. This model is simply a way of organizing tables in a way that makes sense for reporting. For example, the star schema, optimized for analytical reporting. The HANA view is similar to a universe, it allows the end user to connect to it from their BI tool. In the background, the modeler still needs to pull in tables, create joins, leverage primary and foreign keys, etc. So in short, traditional modeling practices and theories still come into play.
2. If extracting and transforming data using SAP's Data Services, the transformation of data will happen in HANA, using the Data Quality Management GUI.
Just as a side note, HANA has 3 methods of extracting
data:
Data Services
SAP Replication
Server
SAP Landscape
Transformation
3. Yes, dimensional modeling is still required. At the end of the day, HANA is still a database. We
haven’t changed the theory of data modeling, it still needs to happen, our approach is just different.
Hope that helps.
Kyle -
Hi Experts,
I am having a requirement to do dimensional datamodel for a datawarehouse for the below scenario. Please provide me your suggestions. How many dimensions and facts which i should consider for this? Given below is my requirements.
Percentage and Numbers of care plans due for review, indate or expired
Analysis by home, keyworker, client
Drill down to home by client
Drill down to keyworker by client
Drill down to client
Source OLTP Tables
Patient : (PatientInfo)
=========
clientName
clientAdress
Age
Staff: (StaffInfo/Keyworker)
============
StaffName
StaffDept
DOB
Careplan:(This table will holds Patient Careplans)(1:M)
=========
Expected Outcome (Expected Result of this actionplan)
Current Situation (Current condition of patient)
Action (what is the action taken for the result)
Start Date (patient start)
review Date (for every care plan there will be review date and on that date this record will expire and create a new record with the current situation)
EndDate (End date of patient)
Keyworker(Staff which is in primary charge for this patient- A staff can be keyworker for multiple patients)
thanks
MUSHi Rajiv,
I'm no expert either and am just beginning to learn HANA, but here are my thoughts (community, please correct me if I am mistaken.):
1. HANA leverages existing data modeling theories, such as the entity relationship model. This model is simply a way of organizing tables in a way that makes sense for reporting. For example, the star schema, optimized for analytical reporting. The HANA view is similar to a universe, it allows the end user to connect to it from their BI tool. In the background, the modeler still needs to pull in tables, create joins, leverage primary and foreign keys, etc. So in short, traditional modeling practices and theories still come into play.
2. If extracting and transforming data using SAP's Data Services, the transformation of data will happen in HANA, using the Data Quality Management GUI.
Just as a side note, HANA has 3 methods of extracting
data:
Data Services
SAP Replication
Server
SAP Landscape
Transformation
3. Yes, dimensional modeling is still required. At the end of the day, HANA is still a database. We
haven’t changed the theory of data modeling, it still needs to happen, our approach is just different.
Hope that helps.
Kyle -
Re: [SunONE-JATO] Re: How to use a tiled view without a model
I'm not sure what is different for you now. You still parse the string
and put it into a data structure. Before the data structure was a
vector, in JATO its just a model with a "hidden" data structure (a hash
map).
MVC only really comes into play when you talk about where your write
this code, and where the data structure is being stored. So really, JATO
takes care of half of the MVC'ness of it all (where the data is store).
You just decide where to be the code to populate the model.
Make sense?
Is there something different required of you in JATO in this scenario
that I am not grasping?
c
Hoskins, John D. wrote:
Thanks for the feedback.
The problem I was solving involved a single string, which contained
delimited subsets of information.
The string looked like
this:"time|analyst|description|time|analyst|description|..."
In ND, I parsed it apart into it's components (time vector, analyst vector,
description vector), populated the repeated.
With JATO, how would I make a model for something that doesn't have a
database component like this?
I'm pretty new to this MVC thing, so bear with me.
John D. Hoskins
Telephone & Data Systems
Application Development & Support
Voice: 608.664.8263
Fax: 608.664.8288
Email: john.hoskins@t...
-----Original Message-----
From: Craig V. Conover [mailto:<a href="/group/SunONE-JATO/post?protectID=219212113009229091025149066024064239039098031198039130252055210">craig.conover@s...</a>]
Sent: 6/26/2002 3.22 PM
Subject: Re: [SunONE-JATO] Re: How to use a tiled view without a model
I guess the only thing "weird" (for lack of a better term) about what
you are doing is that your are populating the model on the "display
cycle". Typically, the cycle goes like this:
Request -> populate model -> update data store -> retrieve data to
populate model -> display data
some of the above steps are optional but hopefully you get the point I
am making.
So what you are doing is:
Request -> populate model/display data
If it works for you, then it's not necessarilly wrong. But I would
probably have my model populated before I forwarded to the target
(displaying view bean) or at a minimum, in the begin display event of
the view bean or the tiled view, but not during the iteration of the
tiled view.
c
jhoskins wrote:
Craig,
Thanks for the pointers. I ended up doing something else. I set the
models setSize() method to set the max size, and as the tiles fields
iterated, populated the value from some vectors I had the data in
already. Is this solution fraught with peril and will ultimately fail,
or should I try your way?
John
--- Craig V. Conover wrote:
John,
Check out the docs for DefaultModel. There is an appendRow() method.
So get your tiledview's primary model (the tiledview's primary model
should be set to use an instance of DefaultModel), model.appendRow(),
then model.setValue("fieldname", value) for each value.
Rinse, repeat as needed.
c
jhoskins wrote:
I would like to use a tiled view, but populate the fields manually.
Any pointers about where I can set the size of the tiled view? I tried
setMaxDisplayTiles() in the beginDisplay, but it won't get down and
generate the rows.
John Hoskins
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jsp
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jsp
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jsp
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jspCraig,
Thanks for the pointers. I ended up doing something else. I set the
models setSize() method to set the max size, and as the tiles fields
iterated, populated the value from some vectors I had the data in
already. Is this solution fraught with peril and will ultimately fail,
or should I try your way?
John
--- "Craig V. Conover" wrote:
John,
Check out the docs for DefaultModel. There is an appendRow() method.
So get your tiledview's primary model (the tiledview's primary model
should be set to use an instance of DefaultModel), model.appendRow(),
then model.setValue("fieldname", value) for each value.
Rinse, repeat as needed.
c
jhoskins wrote:
I would like to use a tiled view, but populate the fields manually.
Any pointers about where I can set the size of the tiled view? I tried
setMaxDisplayTiles() in the beginDisplay, but it won't get down and
generate the rows.
John Hoskins
To download the latest version of JATO, please visit:
http://www.sun.com/software/download/developer/5102.html
For more information about JATO, please visit:
http://developer.iplanet.com/tech/appserver/framework/index.jsp -
How to add an index to a materialized view in Data Modeler 3.3
Hello everyone,
I'm looking for a how-to to add an index to a materialized view in Data Modeler 3.3.0.747, as I coudn't find a way to do it so far.
I looked here:
Relational Model
Physical Model
Oracle 11g
Materialized Views
"my_mv_name"
"INDEXES" IS NOT HERE IN THE TREE
"Tables" does not include it either
Thank you & Best regards,
BlamaHi David,
thanks a lot. I did so and it worked, but I found a minor bug while doing so:
I marked the table as "Implement as Materialized View" and went to File->Export->DDL (for Oracle 11g).
The generated code (I checked all options in "Drop Selection") includes a row:
DROP MATERIALIZED VIEW mv_mymatview CASCADE CONSTRAINTS ;
which produces a syntax error.
Best regards,
Blama -
How to include Abbreviations in Dimensional Model
Dear All,
The question is basically related to modelling but as I am using OWB for this, I am posing the question here.
I have to migrate a HR database from relational model to dimensional model. I have the following situation.
There is a table which has two columns i.e. Country_cod, and country_name like UK- United Kingdom, US - United States. Then in the database where ever it is needed to use information about the country, the country_cod is used. I want to migrate the database to dimensional model, This table does not relate to any other table directly, like a foreign key relationship.
How should I deal with this situation?
Comment please....
regardsYou can put your Javascript code in the Onmouseover property of the column.
I don't have the Javascript code for changing the color but for example the following will pop up an alert
<af:outputText value="#{row.DepartmentId}" onmouseover="alert(\'Unload event fired!\')"> -
Creating a view and use the view in a model
Hi All,
I am new to OBIEE and i have a requirement that i have two fields to be added to a subject area. Since we can't add queries directly to a model, i have to create a view and then put the query in it and then use the view in that model.
I couldn't understand this much, if anybody could help me explaining in detail about the views and adding a query in it?
Thanks
GohanHi,
Please refer the screen short below link
http://imgur.com/b9sXS
Right click any of one physial table(RPD physical layer) -> Select Properties -> from the properties windows choose 'Select' as Table type ->
Please enable feature 'CREATE_VIEW_SUPPORTED' view just go to 'Features' tab and from the available option select 'CREATE_VIEW_SUPPORTED' is checked or not. If it is not checked then check the option.
can you explain your current model and scenario?
for example to you can use to fetch two fields from Subqueries,all trypes of joins(inner,left outer,right outer etc..,) its upto you case
e.x:fetching two tables columns mapping subquery:
SELECT column1, column2 FROM table1 WHERE column1 IN (SELECT column1 FROM table1
INTERSECT SELECT column1 FROM table2 )
SELECT column1, column2 FROM table1,table2 WHERE table1.column1 = table2.column1
Alternative there is feature call "mulitple subject area" method will work (Union,Union all etc..,)
Thanks
Deva
Edited by: Devarasu on Jun 6, 2012 4:40 PM -
Download IBM Book on Star Schema Design (Dimensional Modeling)
Hello Friends,
You can download an IBM Book on Dimensional Modeling from the following link:
http://www.redbooks.ibm.com/redbooks.nsf/e9abd4a2a3406a7f852569de005c909f/e235dc46161249d38525703e00036135?OpenDocument
The Book is very well explained.Dear Krish,
Please try this link and let me know if you can open.If not I will email you the Book:
http://www.redbooks.ibm.com/abstracts/sg247138.html?Open
My Email is [email protected]
Thanks
Dave -
Download Dimensional Modeling Book (Star Schema Design Book)
Hello Friends,
You can download an IBM Book on Dimensional Modeling from the following
link:
http://www.redbooks.ibm.com/abstracts/sg247138.html?Open
This is an excellent IBM Redbook on Data warehouse design and
Dimensional Modeling design.
Great to read,
DaveDear Krish,
Please try this link and let me know if you can open.If not I will email you the Book:
http://www.redbooks.ibm.com/abstracts/sg247138.html?Open
My Email is [email protected]
Thanks
Dave -
Porblem with Dimensional modeling
Hi everybody;
I'm new in datawarehousing.
I want to create a datamart about road accident. I have a very simple E/Rship mode :
drivers
1,1
|
|
|
1,N
|
vehicules 1,1-------1,N-accient-1,N ------- 1,1- victime
|
1,1
|
|
|
1,N
|
cause_accident
My physical model :
table cause_accident(code_cause,description,categrie_cause);
table accident (code_accident*,date_accident,region,number_of_victimes,code_cause**);
table vehicule (matricule*,type_vehicule,code_accident**);
table victime (num_victim*,name,age,sexe,code_accident**);
table driver (num_driver*,name,age,sex,code_accident**);
* :primary key
**:foreing key
So I want to create a datamart about raod accident to be able to get some statistics like :
-number of accident per month, year...(time)
-number of victims per month, year...(time)
-number of accident per region
-number of victims per region
-number of accident per vehicule type
-number of accident per victime sex
-number of accident per victime age
-number of accident per driver sex
-number of accident per driver age
I have the fact table "accident"
dimensional tables : times,region
My question is :
how to have the others dimensional tables because we have 1,N (many to many) relation between accident and: driver,victime and vehicule? Really it's not clear for me at all !!!!
Plz I need for your help .
Thank you at advanceIf you're new to dimensional modeling I'd suggest the following site:
http://www.kimballgroup.com/html/articles.html
Also worth picking up Ralph Kimball's Data Warehousing book.
Without thinking about this too much, maybe the granularity of the Fact table is per Driver/Vehicle, an Accident dimension can then be used to link the rows that were involved in the accident.
Cheers
Si -
Problem with dimensional modeling
Hi everybody;
I'm new in datawarehousing.
I want to create a datamart about road accident. I have a very simple E/Rship mode :
drivers
1,1
|
|
|
1,N
|
vehicules 1,1-------1,N-accient-1,N ------- 1,1- victime
|
1,1
|
|
|
1,N
|
cause_accident
My physical model :
table cause_accident(code_cause,description,categrie_cause);
table accident (code_accident*,date_accident,region,number_of_victimes,code_cause**);
table vehicule (matricule*,type_vehicule,code_accident**);
table victime (num_victim*,name,age,sexe,code_accident**);
table driver (num_driver*,name,age,sex,code_accident**);
* :primary key
**:foreing key
So I want to create a datamart about raod accident to be able to get some statistics like :
-number of accident per month, year...(time)
-number of victims per month, year...(time)
-number of accident per region
-number of victims per region
-number of accident per vehicule type
-number of accident per victime sex
-number of accident per victime age
-number of accident per driver sex
-number of accident per driver age
I have the fact table "accident"
dimensional tables : times,region
My question is :
how to have the others dimensional tables because we have 1,N (many to many) relation between accident and: driver,victime and vehicule? Really it's not clear for me at all !!!!
Plz I need for your help .
Thank you at advanceIf you're new to dimensional modeling I'd suggest the following site:
http://www.kimballgroup.com/html/articles.html
Also worth picking up Ralph Kimball's Data Warehousing book.
Without thinking about this too much, maybe the granularity of the Fact table is per Driver/Vehicle, an Accident dimension can then be used to link the rows that were involved in the accident.
Cheers
Si -
Hello:
I am new to dimensional modeling and trying to design a model based on the existing OLTP system.
Below are the links to ERDs for existing OLTP system and a dimension model that I have developed. Can someone expert in this area take a quick look and offer me suggestions please? It won't take much time as the model is simple. I am especially worried about the inclusion of bridge table STORE_DEPARTMENT in the model and the dimension tables referencing other dimension tables. Is this normal or I am doing it wrong? I am trying to check if the model can answer some of the DSS questions but your suggestions would really help me if I am going in a wrong direction.
Thanks in advance for your help :)
Regards,
RameshHi,
There is a trade off on the availability and the Complex analytics.
A star schema is good if you have the functional requirements really simple. Like the dimension is not SCD Type2 (slowly changing dimension) and you don't need to do "AS IS" vs "AS WAS" reporting.
In modern Analytics in any domain dimensions are SCD Type 2 as business keep on evolving. In a star schema structure this will cause explosion of data if there are frequent changes at the higher levels of the dimensional hierarchy. That anyway will hit the performance.
As far as my experience goes, at the data model level it is better to have snow flaked dimensions. and while managing the metadata (in a BI reporting tool) you can consolidate the snowflaked dimensions in star schema structures. That will make ah hoc analytics much simple for the business users.
A lot of performance measure can be taken to improve the end user experience.
In short the trend in BI analytics demands to have a snowflaked structure rather than a simple star schema structure.
Hope this helps. -
Designing Dimensional Model from Reports
I'm currently desiging a data warehouse for our project. My problem is that we do not have access to the source system. The only thing we are provided with is the data feeds that are generated by the current reporting system. This is causing me a great deal of headach b/c reports do not give you an entire picture of what the source system. Since reports only display data based on how it is developed, it is difficult to come up with an efficient design. My question is has anyone designed a dimensional model using only Reports as the source?
Hi there,
The short answer to your question is 'no'
A longer answer is 'Surely it is possible to get a copy of the source schema definition even if it is not possible to get access to the source system?'
Stuart
Maybe you are looking for
-
Hi Friends! I´m learning ADF with JDeveloper 11g release 1 and i have some problem with the DBSequence! I have done a simple sample to test de ADF operations! I used Fusion MiddleWare template to my application and i have used ADF Business Components
-
Dear Experts, I have created a condition type for material rebate and the same has been assigned to condition type grp in turn condition type grp is assigned to rebatye agreement type....besides the same(cond type) is included in the pricing procedur
-
I can't update my IPhone to version 3.1.3.
Hi, I can't update my software on my Iphone to version 3.1.3. because Itunes says that Itunes is not connect to the internet. I have internet on my mobile and computer. I don't know what to do. I hope that you can help me. Greetings
-
Some easy setup are missing from FCE.
Hi, Im teaching FCE, on some computers I have all the easy setup I need, and on some computers I dont (for example, I miss the Apple intermediate codec 720p 25)... is there a way to install addition easy setup? many thanks
-
Bt rfs date set, for about a few months ago :(
So basically my elderly exchange which i get a measly 7mbps off had its rfs date set for 21CN and adsl2 (woohoo me though, good internet speeds for once!) it was set for: BT have announced a ready for service date for this exchange. The estimated go