Sie sind auf Seite 1von 5

A Business Intelligence Model for Public Sector

Raghav Venkat
Therese Arguelles
City of Las Vegas

Introduction
Public sector agencies are using multiple systems to manage their functions and are data rich. Various
disparate systems collecting data within itself makes it hard to have a unified view of data across the
entire agency. As these systems are usually custom built, there is no out of the box business intelligence
solution. This paper discusses how the power of Oracle Business Intelligence Enterprise Edition can be
harnessed to create a single enterprise system unifying data through advance modeling techniques.
Governments have been using multiple information technology systems to optimize the performance of
its operations. The systems are diverse in nature, independent and self-reliant on resources. They provide
a gold mine of structured and unstructured data that could be put to use in various ways. Any intelligent
enterprise wants to uses a state of the art business intelligence system to mine the data to better
understand, estimate and even predict the outcomes of various operations and new projects. The business
intelligence system should be a rock solid technology which combines many unique capabilities of
adding sense to the various forms and formats of data from independent systems.
There are many off-the-shelf business intelligence systems available for large EPR/CRM applications
currently in the market. These systems enable an enterprise to get a good foundation to address their
business intelligence needs. But, most of the systems used in the public sector are unique in nature or
completely custom built to suit the most detailed and legal requirements. This puts a handicap on
governments to use the off the shelf state of the art technology.
Thus public sector entities rely solely on building a data warehouse from the ground up to meet their
business intelligence needs. This still is a great way to achieve success; in a dynamic and data intensive
world, any technology that is custom build tends to become ineffective or efficient. Most custom built
solutions are also not cost effective and the return on investment metric lags. As public sector projects are
planned to succeed, the technology stack in not the latest or greatest.
This paper proposes a public sector business intelligence model that uses an advance Oracle technology
that is capable of adapting to the dynamic data driven analytics technological advancements. This
proposed model is metadata driven and source agnostic in nature. This methodology gives it a
configurable and reusable property. This approach can also deliver operational and analytical reporting
together without a separate design. The greatest advantage is that this model reduces the total cost of
ownership and improves the time to value of this custom business intelligence environment with
advance state of the art technology stack.
This paper also discusses a model to develop an effective business intelligence and organizational
strategy to succeed in the project and attain data governance.

COLLABORATE 14

Copyright 2014 by Raghav Venkat

Page 1

Unique Public Sector Systems


Most public sector agencies use commercially available enterprise resource planning and Customer
Relationship management systems from top class vendors. In addition to that, public sector business
needs are unique in nature. This makes them stand out of the commercial world to buy off the shelf
applications. The business needs are satisfied by applications that are custom built or on the cloud or
shared between entities or are completely managed by third party vendors etc.
To justify, lets analyze some generic scenarios that most governments might face:
Public Works: This department might use basic, but data intensive systems from plant uses various
SCADA and Laboratory management systems to more advance project management and capital projects
tacking systems.
Fire and Rescue: In most cases a Valley-wide incident response system is used which records enormous
amounts of transactional data at every level of an incident response, from the 911 call center to the fire
department response plan to the time the team returns to the fire station or after providing a health
related service.
Police Department: Systems ranging from inmate management that tracks data relating to booked
inmates, animal control, and crime incident response systems to Parking Enforcement might be used.
These secure systems are the life line of law enforcement and are operational up to the second.
The Assessor, clerk, urban Development etc., all function in conjunction with the GIS applications to
perform its operations. Additionally there are other transactional processing systems for issuing permits
and to enforcing the legal codes.
Its amazing that these systems are designed to capture as much data as possible for a data driven factual
decision making.
The users needed a system that could help them monitor the current data instantaneously, perform data
analytics, report measures, create time lines, and analyze trends and an ability to visualize their data.
Going a step further, the data from these systems when combined together could be more insightful in
predicting trends, finding a hidden truth, relate incidents, and make connections to events and to also
create a data repository that can used to create/generate predictive algorithms.

Technology for the Model


Oracle Business Intelligence Foundation Suite (OBIEE) is a complete, open, and architecturally unified
business intelligence system for the enterprise that delivers abilities for reporting, ad hoc query and
analysis, online analytical processing (OLAP), dashboards, and scorecards. All enterprise data sources, as
well as metrics, calculations, definitions and hierarchies are managed in a Common Enterprise
Information Model, providing users with accurate and consistent insight, regardless of where the
information is consumed. Users can access and interact with information in multiple ways, including
web-based interactive dashboards, collaboration workspaces, search bars, enterprise resource planning
(ERP) and customer relationship management (CRM) applications, mobile devices, and Microsoft Office
applications.

COLLABORATE 14

Copyright 2014 by Raghav Venkat

Page 2

Important Components
Common Enterprise Information Model: The semantic model of OBIEE. It is accessed via an open API,
making it available to any Oracle or non-Oracle delivery channel, thus providing a common version of
the truth for all Business Intelligence users and applications
Oracle BI Server: A highly scalable, highly efficient query and analysis server that integrates data via
sophisticated query federation capabilities from multiple relational, unstructured, OLAP, and prepackaged application sources, whether Oracle or non-Oracle.
Ad hoc Query and Reporting: A powerful ad-hoc query and analysis environment that works against a
logical view of information from multiple data sources in a pure web environment. This single interface is
designed to seamlessly handle both relational and OLAP style analysis.
Interactive Dashboards: Rich, interactive pure Web dashboards that display personalized information to
help guide users in effective decision making.

The Model
The Oracle BI Foundation Suite allows an organization to model the complex information sources of their
business as a simple, semantically unified, logical business model. It provides facilities to map complex
physical data structures including tables, derived measures, and OLAP cubes into business terms abstracting how a business user expresses calculations. It translates familiar, easy-to-understand business
concepts into the technical details required to access the information. The Oracle BI Foundation Suite is
unique in the market because it defines an enterprise semantic layer that spans across the unified
enterprise view of information.
Taking advantage of this technology, the model designed is completely based on the semantic layer of the
OBIEE suite.
The metadata model is designed logically to address the overall business needs such as metrics, calculations, data
federation and factual analysis of the data independent of the disparate sources that the data is sourced from.
The most expensive part of any business intelligence implementation is the cost associated with the
extract transform and load of the data from source to a staging to a target data warehouse. This also
makes the BI design rigid; meaning the cost of changes to any business requirement is expensive as well.
This method also has some limitations when integrating data between disparate sources. After all the
data warehouse cannot provide operational intelligence, but probably a tweaked system can provide a
mid-latency reporting. An improper data warehouses can causes the BI project to fail!
For this reason, this proposed model would minimize the use of a warehouse to only objects that help in
aggregation of selected data and performance. As discussed about this model can be source agnostic
when creatively designed.

COLLABORATE 14

Copyright 2014 by Raghav Venkat

Page 3

Business Model and Mappings


This layer acts as the core of the proposed model. This layer is modelled the way the business elements
function. Define all the business metrics those are specified in the user requirements. Irrespective of the
source, logically model a dimensional structure that the requirement commands. Create basic facts,
dimensions and hierarchies that you many need. They might come from various sources, sometimes
multiple sources and might have different structure physically. The Logical model enables you to choose
multiple sources and combine them to make it into a dimensional object.
Model the conformed dimensions and hierarchies, measures (including aggregation rules, complex
business calculations, dimensionality and time series), data security rules, and human readable attributes
and dictionary definitions. The mappings from the semantic objects back to the physical objects define the
federation and aggregate navigation across multiple sources. Because of this layering and mapping, the
physical source can migrate to a different brand of database, or even add an aggregate, without
impacting the business model, presentation layer or reports.
For example, consider the Fire and Rescue departments metric: Response times. This metric can be
logically modeled with facts such as response, time and dimensions such as battalion, location, time,
incident, fire station, command etc. There could be a combination of multiple physical tables that make
up these dimensions and facts. It could be sources from disparate applications with varied data
structures. The logical table sources option will enable you to map this dimensional model to the
physical source.
The presentation will explain this in detail with a specific case.
Once this logical dimensional design is complete, you can design the presentation layer objects. Knowing
the knowledge level of users design the presentation layer objects sourced from your logical model. This
could combine multiple facts under a similar business unit or the same fact or dimension can span across
multiple business units.

How does this work?


The OBIEE server performs two major functions. First it compiles incoming query requests into
executable code, and secondly, it executes that code. In the first step the Oracle BI server performs
various functions such as query parsing, generating a logical request, navigation, rewrites and code
generation. So the metadata we designed in the logical layer combines the business logic and the sources
where the server needs to look for the data. This code generated is the executable code. The execution
engine then executes the code in parallel. The BI server is quite unique in parsing and compilation
techniques; content aware data federation; parallel execution; connectivity adapters; custom memory
management and latch contention.
A more detailed case will be explained in the presentation.
The presentation layer sourced through Common Enterprise Information Model eliminates the need for
business users to understand physical data storage and enables them to combine data from multiple
enterprise information sources quickly and easily.

COLLABORATE 14

Copyright 2014 by Raghav Venkat

Page 4

The Analyses and Dashboards feature that is available with OBIEE provides end users with broad ad-hoc
query, analysis and reporting capabilities in addition to picture perfect dashboards and portal pages. This
being web based is easy to access and use. Users interact with a logical view of the information
completely sheltered from data structure complexity. Users can also easily create a range of interactive
content types which can be saved, shared, modified, formatted, or embedded in the users personalized
dashboard or enterprise portal.

Project BI Strategy
The best way to navigate a project with such models is to develop an effective strategy which is complete,
provides solutions to problems and drives the project to success.
As every business intelligence need is very dynamic, strict governance procedures are to be set. Iterations
have the power to successfully satisfy demands that arise at different points of time. Plan your strategy to
involve multiple short stints or iterations. In this approach all iterations should have specific goals whose
outputs would be acting as inputs to the next. These iterations could be phases for every business unit to
structure their required metrics and implement them independent of others. Do included phases for
integrating the independent phases to one large effective intertwined enterprise wide solution. This
tightly integrated and cyclical approach will eventually prove very productive and successful. This
approach also will also be useful in scheduling timelines for the project. Schedules, a function of cost
estimates and budget, with tight deadlines should be planned. A strategic change control plan and risk
mitigating plan should also be penned. A plan for constant monitoring of the projects progress is
important.

Conclusion
Thus, by enabling the power of the common enterprise information model of Oracle business intelligence,
we are able to successfully create a business intelligence model suitable for enterprise wide use
combining disparate data sources and structures in the public sector to give an intuitive view of their
greatest asset- DATA.

COLLABORATE 14

Copyright 2014 by Raghav Venkat

Page 5

Das könnte Ihnen auch gefallen