After a Very long time , I had some time to Check for some blogs which i was thinking to Jot down into my blog for further reference
Oracle Business Intelligence - an overview
At work I have to deal with OBIEE ( Oracle Business Intelligence
Enterprise Edition ) as a
developer and administrator, so I decided to share the amount of knowledge that
have accumulated. I hope it
will help someone to learn the system.First, I plan to talk about the system as
a whole, then - about the architecture and interaction of the components, and
in the end - to tell you about those features encountered during the work.
System
Overview
Oracle Business Intelligence Enterprise Edition (OBIEE, OBI) - a software platform for solving business analytics:
Interactive and published reports, KPI monitoring, and business processes. Is a descendant of Siebel Analytics. Main
functional parts are available to users:
- Answers
(also called Interactive Dashboards) - build
interactive reports available to users through a web interface OBIEE. Reporting unit
is the analysis - a simple report usually includes one map (table, graph,
chart, etc.). Analysis
combined into dashboards (Dashboards), they work with end users. Dashboards can
also include invitations (prompts) - setting elements with which the user
can interact with the panel.
- Publisher
(sometimes it include Delivers) - a means of
creating and distributing static reports. Because publisher
evolved from a single product, then it has the opportunity to build their
own reports on the data model, independent of the data model OBI, but you
can use a common model. Also there is
the possibility to send existing reports of answers.
- Action
Framework - a set of tools
to perform any automated action from OBIEE - such as sending reports to
achieve a certain measure of value, call external Web services, call java
code, etc. Actions can be
grouped into chains.Possible to set up different channels alert users -
ee-mail, sms, pager, etc.
- Scorecard
and Strategy Management - a means to
track key performance metrics (KPI) and working with scorecards (Scorecard
- http://en.wikipedia.org/wiki/Balanced_scorecard ). Used to visually
display the status of implementation of goals (company, project). User access, as
is the case with the answers, through a web interface.
- Marketing - integration
with Siebel Marketing.
- Office
Tools - integration
with Microsoft Office.
Architecture
OBIEE, in fact, is the mediator between the data
sources and users (user reports). The
system does not store the data (excluding caches) - all reports are built
"on the fly" by querying data sources (pictures from the book OBI 11g
Developer's Guide):
OBIEE supports a variety of data sources,
including relational databases, systems OLAP, files and apache hadoop. The system allows you to combine in one report data from different
sources, combining them with each other according to the rules.
At the heart of the system is a common data model (also called repository) - description of the logical model of the business area and bind it to the physical data sources. The model consists of three layers - presentation, business, and physical.The business logic layer is described by the model (which is understandable to end-users), in the form of a multi-dimensional model - the facts and measurements, and describes the binding of logical attributes to physical sources.Presentation allows you to split a logical model for several domains and restrict user access to various parameters and attributes. The physical layer describes the data sources - tables, columns, keys, data cubes. Creating a data model (repository) performs developer using special software - Oracle BI Administration Tool.
When a user opens a report, presentation server (Presentation Server) generates a query using Logical SQL server BI. BI server parses the request and translates it into queries to data sources in their "native" languages - sql, mdx, etc. After receiving data from source server combines them, conducts various operations on the data (for example, calculates the units if necessary), and returns the result to the server presentations. Presentation server, in turn, renders the data in the web interface or generate static report.
At the heart of the system is a common data model (also called repository) - description of the logical model of the business area and bind it to the physical data sources. The model consists of three layers - presentation, business, and physical.The business logic layer is described by the model (which is understandable to end-users), in the form of a multi-dimensional model - the facts and measurements, and describes the binding of logical attributes to physical sources.Presentation allows you to split a logical model for several domains and restrict user access to various parameters and attributes. The physical layer describes the data sources - tables, columns, keys, data cubes. Creating a data model (repository) performs developer using special software - Oracle BI Administration Tool.
When a user opens a report, presentation server (Presentation Server) generates a query using Logical SQL server BI. BI server parses the request and translates it into queries to data sources in their "native" languages - sql, mdx, etc. After receiving data from source server combines them, conducts various operations on the data (for example, calculates the units if necessary), and returns the result to the server presentations. Presentation server, in turn, renders the data in the web interface or generate static report.
The server part of OBIEE includes several
disparate components, some of which is controlled by the application server
Weblogic, and the other part there as usual (native) program and managed
component oracle process manager (opmn).Also OBIEE uses to store data of the
official relational database (Oracle, MS SQL, IBM DB2, or MySQL). OBIEE domain management via a web-interface Weblogic
Administration Console and Fusion Middleware Enterprise Management Console.
How
does the system work
Suppose there is a need of business users to
receive analytical information about the status of any area of the company (for example, sales
volumes in riteylingovoy chain stores). How to
meet the needs of users with OBIEE? Steps
will be approximately as follows (in this case an interactive reports Answers):
1. Determination
of the available data to build reports
have business may already be trained and active data repository for consistent and consolidated data. Then the task easier - we take the data from an existing shop and directly appeal to them of OBIEE. Another thing - if the CD is not available, and there are, for example, data from multiple disparate systems. Typically, these data are in normalized form (NF 3 or higher), and very detailed, which is not good for building reports. In this case, you must design and build a data mart schema type "star" or "snowflake" (or more, depending on the complexity of the data).Showcase can be implemented as tables, or for example, the representations (conventional or materialized).Naturally, this requires constantly working database, capable of withstanding a sufficiently large number of requests.
also possible that the data is available in the form of OLAP - then, as a rule, no revision is not required because this means that the multidimensional data model has been built and operated.
have business may already be trained and active data repository for consistent and consolidated data. Then the task easier - we take the data from an existing shop and directly appeal to them of OBIEE. Another thing - if the CD is not available, and there are, for example, data from multiple disparate systems. Typically, these data are in normalized form (NF 3 or higher), and very detailed, which is not good for building reports. In this case, you must design and build a data mart schema type "star" or "snowflake" (or more, depending on the complexity of the data).Showcase can be implemented as tables, or for example, the representations (conventional or materialized).Naturally, this requires constantly working database, capable of withstanding a sufficiently large number of requests.
also possible that the data is available in the form of OLAP - then, as a rule, no revision is not required because this means that the multidimensional data model has been built and operated.
2. Building
a data model
must describe the existing data and bind logical attributes (eg, sales metrics) from the physical attributes in the database or server OLAP. Based on these data constructed multivariate model - description of the data in terms of facts, dimensions, attributes, dimensions and hierarchies.
must describe the existing data and bind logical attributes (eg, sales metrics) from the physical attributes in the database or server OLAP. Based on these data constructed multivariate model - description of the data in terms of facts, dimensions, attributes, dimensions and hierarchies.
3. Creating
a repository
now developed a data model should be moved to the repository OBIEE. This is done in the Oracle BI Administration Tool (mentioned just above). Development takes place in three stages - import metadata sources, the construction of the business layer and the presentation layer construction. As a rule, there is only one data source, but there may be more complex scenarios, such as combining data from a relational database and OLAP, or merge data with different levels of granularity from multiple databases. In these cases, the developer must also correctly configure the "relationship" between the sources. Building the business layer is mainly composed of the attribute transfer physical layer description hierarchies, select the types of aggregations of metrics and setting tables for logical sources. Presentation layer usually is a reflection of "1 1" business layer, sometimes broken into separate areas (for example, if want to share access to data for users). Also worth noting is that OBIEE has some funds for the joint development of the repository - there is a possibility of their merging of different versions, and can be stored in the repository as a set of xml files for the convenience of working with version control systems.
now developed a data model should be moved to the repository OBIEE. This is done in the Oracle BI Administration Tool (mentioned just above). Development takes place in three stages - import metadata sources, the construction of the business layer and the presentation layer construction. As a rule, there is only one data source, but there may be more complex scenarios, such as combining data from a relational database and OLAP, or merge data with different levels of granularity from multiple databases. In these cases, the developer must also correctly configure the "relationship" between the sources. Building the business layer is mainly composed of the attribute transfer physical layer description hierarchies, select the types of aggregations of metrics and setting tables for logical sources. Presentation layer usually is a reflection of "1 1" business layer, sometimes broken into separate areas (for example, if want to share access to data for users). Also worth noting is that OBIEE has some funds for the joint development of the repository - there is a possibility of their merging of different versions, and can be stored in the repository as a set of xml files for the convenience of working with version control systems.
4. Designing
Reports
After the formation of the repository and upload it to the server begins main phase - development of reports. First developed separate analyzes, then they are integrated into the information panel. As a rule, each developer is working on its own set of analyzes and dashboards, as funds for joint development is not (all happening in the web interface), and simultaneous editing one will overwrite the work of another :)
After the formation of the repository and upload it to the server begins main phase - development of reports. First developed separate analyzes, then they are integrated into the information panel. As a rule, each developer is working on its own set of analyzes and dashboards, as funds for joint development is not (all happening in the web interface), and simultaneous editing one will overwrite the work of another :)
A
little bit of personal experience
Here I have tried to describe some features of
the system that caused the greatest frustration in the development. The list is far from complete, but may be useful in the planning
of the system architecture.
- OBIEE
not always work correctly with schemes such as "snowflake" in
the data model. This means that
not always generates the correct SQL query report. As possible to
translate such a scheme into a "star" at the business layer.For
example, if there is a table "Customer", which refers to a table
"class customer" (individual, corporate client, etc.), the
business layer is necessary to reduce them to a single logical table
"Customer" with a full set of attributes .The situation becomes
more complicated when there is communication fact tables through several
dimension tables.In such cases it is necessary to monitor the correctness
of query generation.
- In
OBIEE have the option of drawing up analyzes on the basis of direct
queries to the database. This requires a
change in the configuration file NQSconfig.INI. This feature
often makes life easier if you need to implement an artful display logic,
without complicating the model without the need for data. However, in this
case, you have to remember which data should have access, business users,
and does not give the opportunity to query the database all in a row.
- You
have to properly configure caching data tables. In that case, if
the data changes are planned during the working day, that people need to
see in the reports - you must either disable caching, or manually (via
WLST) to update the cache. Also good
practice to "warm up" (seeding) cache before workday users so
that users can immediately use all reports.
- It
should be remembered that the functional dashboards as a Web application
is very limited. If you want a
serious interactivity for users to interact with the interface, it is
better to look towards other BI tools such as stack MS. All that can be
obtained in OBIEE - this filter selection, work with data in tables
(sorting order of the columns, creating groups, adding the results and the
like) and so-called "master-detail" events - when the user can
click on the cell in the same table, and in the next graph or table data is
automatically filtered by the selected value in the cell. A functional
action (actions), but he also is very limited - there is only a transition
to a URL, call REST method and transition to other information panel.
Mobile
Business Intelligence
In the arsenal of OBIEE also have the means to
mobile analytics (using mobile devices):
1. Oracle BI
Mobile - app for
ios, which allows you to view the content of dashboards and analysis on the
mobile device. The
mapping is performed almost without changing the appearance of the report, why
it looks a bit in the style of 90-x :)
2. Oracle BI
Moblie App Designer -
application that integrates in OBIEE , you can create reports on
HTML5 using the data model of OBIEE or Publisher. In fact, it is the generator of web applications - each report
consists of several pages, with interactivity and transitions between them. The advantage of this solution is that the application will look
the same in the full browser and on the device, and no need to install a
separate application. Negative
- that the data is not cached on the device, respectively, to use it only if
you have an Internet connection. Mobile
App Designer was released recently and is not yet included in the basic
delivery OBIEE.
Posts
