Description
In the 21st century, organizations are experiencing environmental changes characterized by indistinct organizational boundaries and fast-paced change.
Enhanced Business Intelligence - Supporting Business
Processes with Real-Time Business Analytics
Andreas Seufert
University of Applied Science Ludwigshafen
[email protected]
Josef Schiefer
Institute for Software Technology
and Interactive Systems
[email protected]
Abstract
In the 21st century, organizations are experiencing
environmental changes characterized by indistinct
organizational boundaries and fast-paced change. As a
result firms need appropriate decision support
infrastructures in order to face these challenges. Current
Data Warehousing and Business Intelligence approaches
are widely accepted as a middleware layer for state-of–
the-art decision support. However, they do not provide
sufficient support in dealing with the upcoming
challenges, such as real-time and closed loop decision
making. In this paper, we suggest an architecture for
enhanced Business Intelligence that aims to increase the
value of Business Intelligence by reducing action time
and interlinking business processes into decision making.
1. Introduction
In the 21st century, organizations are evolving into new
forms based on knowledge and networks in response to an
environment characterized by indistinct organizational
boundaries and fast-paced change. Firms are experiencing
environmental changes resulting from the new economics
of information [10] and the increasingly dynamic and
global nature of competition [5] . Therefore organizational
survival depends on the construction and integration of
knowledge fostering the adaptation to the environment, as
well as stimulating environmental changes through the
firm’s knowledge and practices [8]. As a result especially
investments in IT that enable differentiation are of ever
increasing importance [19].
An important component of this investment is in
Business Intelligence (BI). Coined by Gartner in the early
1990s, the term BI denotes on the one hand an analytic
process that transforms internal and external data into
information about capabilities, market positions, activities,
and goals that the company should pursue in order to stay
competitive. On the other hand, BI stands for Information
System concepts like Online Analytical processing
(OLAP), Querying and Reporting, or Data Mining that
provide different methods for a flexible goal-driven
analysis of business data, provided through a central data
pool.
BI may facilitate the connections in the new-form
organization, bringing real-time information to centralized
repositories and support analytics that can be exploited at
every horizontal and vertical level within and outside the
firm [20].
This research aims at contributing to the future
development of BI. It is grounded in the working
assumptions of the information processing theory [29]. As
BI is a newer form of decision support systems [9], [28]
we build on Simon’s [27] perspective of decision making
and Weiner’s classic model of an organization as an
adaptive system [30]. Therefore, we conceptualize inputs
being processed into outputs which feed back to influence
inputs and enable adaptation to external uncertainty.
The remainder of this paper is structured as follows:
Section 2 reviews related work and describes major future
challenges for decision support and Business Intelligence.
Section 3 introduces a framework for enhanced Business
Intelligence integrating the process-oriented and real-time
analytics perspective. In Section 4-5 we describe an
architecture for enhancing Business Intelligence and its
contribution for advanced decision support.
2. Related Work
Due to the challenges in the competitive environment,
two major challenges of the traditional business
intelligence (BI) concept can be identified.
Convergence of Business Processes and BI
First, the common thread running through the new-
form organization is that the organization’s foundation is
the value chain – a set of primary and secondary activities
that create value for customers [7]. Therefore, BI is
targeted to support process-oriented organizations.
Initiated by the works of Hammer and Champy [14],
Davenport [6] and others, companies have redesigned
their organizations around business processes. With
process orientation gaining importance, the need for
effectively managing and controlling business processes is
of great importance resulting in new requirements for
decision support. Collecting and reconciling all
operational data related to business processes, enables the
measurement of process performance and helps to identify
opportunities for process improvement.
As management decisions require the integration of
decision-relevant information, operational data from
different applications have to be collected, integrated and
prepared for data analysis. Today, this is mainly achieved
by using traditional data warehouse systems and
traditional business intelligence tools [31]. In most cases,
the resulting management information systems are not
primarily targeted on measuring the performance of
business processes but on fulfilling traditional reporting
requirements (e.g. financial reporting) [12].
Convergence of BI and EAI
Second, Business Intelligence aims at providing a
closed-loop support that interlinks strategy formulation,
process design and execution with business intelligence
[11], [22].
In order to increase their competitiveness, companies
strive towards reducing the time needed to react to
relevant business events. An ideal state would be reached
if reactions were possible in real-time, i.e. without any
latency between recognizing a relevant business event and
taking an appropriate action. Enterprise Application
Integration (EAI) suites provide a popular solution for
integrating heterogeneous applications in or near real-time
because they are able to seamlessly publish any kind of
data updates to every subscribing application.
However, this integration usually is achieved between
operational systems and involves only little data
consolidation. When it comes to extensive data analysis,
Business Intelligence will be used to produce the
information that is necessary to decide and take
appropriate actions. Addressing this field, real-time
decision support gained great attention. Concepts such as
active warehousing [3], real-time analytics [24], real-time
warehousing Error! Reference source not found., or
real-time decision support provide suggestions of how to
speed up the flow of information in order to achieve
competitive advantage.
Additionally, the vendors of BI and data warehousing
solutions tend to enhance their products by mechanisms
for real-time data integration and real-time analysis: this
leads to the convergence of EAI and BI solutions [21].
3. Framework for enhanced BI
BI as Decision Support Middleware Layer
Data warehouse systems are widely accepted as a new
middleware layer between transactional applications and
decision support applications, thereby decoupling systems
tailored to an efficient handling of business transactions
from systems tailored to an efficient support of business
decisions. The transformation of transaction data into
integrated, consistent input for decision support
applications by data warehousing consumes a certain
amount of time and creates non-volatile, aggregate
information. Many operational decisions (e.g. promotion
effectiveness, customer retention, key account information
[17], need actual yet integrated and subject-oriented data
in or near real-time [26].
Basic orientation: Transaction Subject
Time references: Actual Actual + historical
Access: Read-write Read-only
Aggregation level: Detailed Aggregated
Integration level: Isolated Integrated
Accessibility: Real-time Delayed
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Basic orientation: Transaction Subject
Time references: Actual Actual + historical
Access: Read-write Read-only
Aggregation level: Detailed Aggregated
Integration level: Isolated Integrated
Accessibility: Real-time Delayed
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Figure 1: Operational data stores vs. Data Warehouse
(adapted from Winter [32])
Therefore, the concept of operational data stores has
been introduced for operational decision support [18] and
for data-oriented application integration [15]. It becomes
evident that operational data stores can be positioned
between transactional applications and the data warehouse
[17], [18]. Data managed by operational data stores have
characteristics that differ from data managed by
operational applications, as well as from data managed by
the data warehouse (see figure 1).
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
Vertical
applications
Cross-
product
applications
Operational
data stores
Data staging
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Channel oriented
integration
Product oriented
Integration
Core data
Integration
Business
function
Business
process
Business
unit
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
Vertical
applications
Cross-
product
applications
Operational
data stores
Data staging
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Channel oriented
integration
Product oriented
Integration
Core data
Integration
Business
function
Business
process
Business
unit
Figure 2: Business Intelligence as Decision Support
Middleware Layer (adapted from Winter [32])
Since both the data warehouse and operational data
stores can be regarded as data-oriented integration
middleware, it was proposed that operational data stores
should be implemented as a part of the data warehouse, or
that the data warehouse should be directly utilized for
transactional services like customer relationship
management or e-commerce [23]. Since operational data
stores are fundamentally different from the data
warehouse due to real-time processing needs and read-
write access to data, these two middleware layers are
usually separated in the application architecture.
Therefore, Winter suggests the application architecture for
decision support [32] shown in Figure 2.
Enhancing BI with Real-Time Business Analytics
Data warehousing may contribute to an efficient
information supply between transactional applications and
decision support applications. However, information
‘backflows’ take their way indirectly (i.e. by actions of
decision makers) into the transactional systems and not
directly through the data warehouse (open-loop approach).
Operational data stores, in contrast allow an efficient
‘local’ closed-loop supported between vertical and
horizontal transactional applications.
What is decisive is a closed loop approach interlinking
operational and strategic decision making. Therefore, BI
has to be enhanced towards a closed loop real-time BI,
shortening the period of time between the occurrence of a
business event that requires an appropriate action by the
organization and the time the action is finally carried out.
According to Hackathorn [13], the additional business
value of an action decreases, the more time elapses from
the occurrence of the event to taking action.
d
a
ta
la
te
n
c
y
Figure 3: Business value and reduced action time
(adapted from Hackathorn [13])
The elapsed time is called action time and can be seen
as the latency of an action. Action time comprises four
components:
1. Data latency is the time from the occurrence of the
business event until the data is stored and ready for
analysis.
2. The time from the point when data is available for
analysis to the time when information is generated out
of it is called analysis latency; it includes the time to
determine root causes of business situations.
3. Decision latency is the time it takes from the delivery
of the information to selecting a strategy in order to
change the business environment. This type of latency
mostly depends on the time the decision makers need to
decide on the most appropriate actions for a response
to the business environment.
4. Response latency is the time needed to take an action
based on the decision made and to monitor its outcome.
This includes communicating the decision made as a
command or suggestion, or executing a business action
in a target system.
4. Architecture for Enhancing BI
In this section, we propose an architecture for real-time
analytics with the aim of reducing the action time and
thereby increasing the value of Business Intelligence.
Figure 4 shows an architectural diagram for real-time
analytics with two infrastructure types: 1) information
integration infrastructure and 2) business integration
infrastructure. The main objective of this architecture is to
seamlessly integrate the two infrastructure types in order
to minimize the aforementioned latencies. In the diagram
we extended the components and modules of traditional
BI in order to enable real-time analytics for business
environments.
The information integration infrastructure is
responsible for managing the data for business intelligence
purposes and offers data analysis to decision makers and
to IT systems. Traditional Business Intelligence aims to
support strategic decision makers and therefore uses
analytical applications that are periodically fed with data
from the data warehouse. These analytical applications are
generally completely disconnected from operational IT
systems. Decisions are executed by communicating them
as a command or suggestion to humans. On the other
hand, the enhanced Business Intelligence includes
analytical services which are continuously fed with data
from the operational environment (e.g. via the ODS) and
can be directly invoked by other systems. The object of
analytical services is to provide continuous data analysis
that is able to also cope with current changes in the
business environment.
The central piece of the Business Integration
infrastructure is a Sense & Respond (S&R) system that
communicates events via hubs with the internal and
external business environment. Table 1 provides an
overview of the modules of the S&R system. The internal
business environment comprises vertical and horizontal
applications which are shown on the left side of the
diagram. From the external business environment on the
right side, events are captured during the collaboration
with business partners, or when the contents of websites
changes (e.g. a competitor updates the product prices).
Information Integration Information Integration
Infrastructure Infrastructure
Analytical Services Analytical Services
Sense & Sense & Respond Respond
Horizontal Horizontal
Applications Applications
Horizontal Horizontal
Applications Applications
Business Integration Business Integration
Infrastructure Infrastructure
Analytical Processing Analytical Processing
Decision Making Decision Making
Response Management Response Management
Situation Detection Situation Detection
Event Transformation Event Transformation
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
Events
Metrics
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
H
u
b
H
u
b
H
u
b
H
u
b
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
Events
Metrics
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
OLAP / OLAP / Data Data Mining Mining
Data Data Warehouse Warehouse / ODS / ODS
Traditional Business Intelligence Enhanced Business Intelligence
Vertical Vertical
Applications Applications
Vertical Vertical
Applications Applications
Analytical Applications Analytical Applications
H
u
b
H
u
b
H
u
b
H
u
b
Web Web
Information Information
Collaboration with Collaboration with
Business Partners Business Partners
Figure 4: Architecture with enhanced Business Intelligence
Module Description
Event
Transformation
Transformation of the captured events from the
business environment into meaningful business
information, such as key performance indicators.
Situation Discovery Detection of business situations and exceptions in
the event data and business information. For
instance, an organization wants to detect
suspicious customer behavior (e.g. fraud behavior)
which should be countered with a proactive
response.
Analytical
Processing
Invocation of analytical services in order to
determine the root causes for business situations
and exceptions. Also, prediction of the
performance and assessment of the risks for
changing the business environment. Please note,
that the S&R system delegates the analytical
processing to the analytical services of the
information integration infrastructure. The S&R
system correlates and collects event data for the
data analysis, invokes the analytical services and
processes the results.
Decision Making Based on the analytical results, selection of the
best option for improving the current business
situations and determining the most appropriate
action for a response to the business environment.
This step can be automated with rules, or done by
involving humans.
Response
Management
Response to business environment by
communicating the decision made as a command
or suggestion (e.g. by e-mail), or by directly
adapting and reconfiguring business processes and
IT systems.
Table 1. Modules of the S&R System
During the event processing in the S&R system,
business information is continuously generated and
decisions are made to which a response follows. The
response has an effect on the source systems (from which
the S&R system originally received the events) and
consequently also on the performance and the success of
the organization. In order to reduce action time, latency
has to be reduced in all four stages of the action time. In
the following we discuss how our extensions to traditional
Business Intelligence help to minimize the various types
of latencies.
Minimizing Data Latency
Data Latency is usually mainly influenced by the
refresh cycle of the data warehouse system in which the
data is stored for analysis purposes. A main drawback of
the data warehouse concept is the time-consuming and
resource-intensive process of extracting data from
operational systems, transforming it and loading it into the
data warehouse database. Due to this fact, the so-called
ETL processing is often executed in batch mode at non-
peak times (e.g. overnight), causing time-lags between the
recognition of a business event and its delivery for
analysis purposes.
In contrast to the data warehouse system that requires
extensive data cleansing, data consolidation and data
quality management, an ODS stores a limited scope of
data with only basic (or even none) consolidation and data
quality management, thereby allowing real-time or near
real-time updates and faster data distribution [16].
Another way of reducing data latency is to change from
a periodic batch-oriented to an event-driven update of the
data warehouse by using EAI technology. By doing this,
data representing a certain business event will be
populated into the data warehouse as soon as the event is
recognized in an operational system [25].
We extend the approach from Schiefer and Bruckner
[25] by integrating event streams from the business
environment with integration hubs. The integration hubs
unify the raw event data and feed them continuously into
the S&R system. The S&R system generates business
metrics that are stored in a real-time data store (e.g. an
ODS) and used as input for the invocation of analytical
services.
Minimizing Analysis Latency
Analysis latency is mainly determined by the time it
takes to inform the person in charge of data analysis that
new data has to be analyzed, the time needed to choose
appropriate analysis models and the time to process the
data and present the results. Current approaches that deal
with the reduction of analysis latency are provided by BI
software. A prominent BI concept is Online Analytical
processing (OLAP), which concentrates on reducing the
time needed for analyzing the data by providing powerful
user-interfaces that let the analyst explore the data along
previously defined analysis dimensions [4].
In contrast to the fixed dimensional structures
necessary for doing OLAP, Data Mining is a more flexible
BI approach which allows the application of different data
exploration techniques to a large amount of data in order
to discover unknown relationships between variables or
single data items [1]. While pure data mining is mainly
focused on reducing the time for data processing by
applying efficient data exploration algorithms, data
mining software tools like SPSS or IBM intelligent miner
also facilitate the process of selecting and adapting a data
mining technique that is suitable for a given problem. In
order to reduce the time for notifying the analyst of
business events, the approach of automated exception
reporting can be used.
In order to reduce the analysis latency, we use in our
architecture analytical services in the information
integration infrastructure which enable an invocation of
analytical function without manual intervention. A
service-oriented architecture for analysis provides
standardization for using various types of analysis
techniques within a BI process. Unlike traditional BI
tools, analysis services are permanently available, which
is a prerequisite for a continuous and automated BI
process.
Minimizing Decision Latency
The least IT support can be identified in the area of
decision latency. In most cases, the interpretation of
analysis results and the derivation of appropriate actions
are seen as manual processes that have to be carried out
by knowledge workers and are therefore time consuming.
Newest advances especially in the area of Business
Activity Monitoring (BAM) try to improve this situation
by automating certain decision processes with the help of
rule-based decision engines. Based on the real-time
analysis of data from an EAI platform, the decision engine
checks for predefined business rules and notifies
responsible people, or triggers other tools for conducting
further actions.
In our architecture we follow the approach from BAM
and use rule-based decision making for automating many
operational and tactical decisions. Please note, that the
rules for these decisions are in many cases derived from
strategic decisions that are made by humans. The decision
rules help to intelligently respond very quickly to the
current business situation.
Minimizing Response Latency
The final outcomes of a Business Intelligence process
are actions based on the decisions made (manually or
automated). There are two kinds of response latencies: 1)
the time it takes to initiate an action and 2) the time it
takes to execute and monitor the action. Hackathorn
considers in [13] the response latency as part of the
decision latency which is reasonable for strategic decision
making. When it comes to real-time analytics, it is also
crucial to communicate and carry out a decision very
quickly. We consider separately in our model the time it
takes to initiate and execute an action in the business
environment. For this reason, a S&R system includes a
response management module which is responsible for
triggering business operations and monitoring their
outcome.
The business value of our architecture with enhanced
Business Intelligence can be summarized as follows:
- Real-time business information: Minimal latency for
preparing and analyzing data and hence improved
visibility and accuracy of business performance
indicators. The indicators are also available for
operational and tactical decisions.
- Optimized business processes: By integrating real-
time analytics, the internal and external business
environment can be optimized by more efficient and
intelligent control mechanisms. The S&R system
operates as an “external advisor” for the business
environment.
- Automatic discovery of situations and exceptions:
A S&R system supports a continuous discovery of
business opportunities and exceptions based on the
current state of the business environment. For instance,
companies are able to detect suspicious customer
behavior (e.g. fraud behavior) which can be countered
with a proactive response.
- Proactive responses: By continuously observing and
analyzing customers, business partners and the
competition, the business environment can be
proactively adapted and optimized.
- Generating more accurate forecasts in near real-
time under consideration of current and historic data
(e.g. continuous update and optimization of production
plans based on the current orders).
- Integrating internal and external source systems:
The S&R system is able to correlate and merge event
streams from the internal and external business
environment.
- Less integration effort: The S&R system has a
significantly lower integration effort than traditional
data warehouse solutions since only events from the
source system have to be integrated. For the event
processing the S&R system does not have to know
internal details of operational systems, such as the data
model.
5. Real-Time Business Analytics in Action
Let’s assume that the company Meyer AG offers
transportation services for various goods. In order to
better monitor, synchronize and optimize the processing
flows, the company uses mobile devices (e.g. mobile bar
code scanners) to store information for every article about
the current location and the status of the current method of
transportation. The S&R system continuously processes
and analyses this data and calculates indicators which
provide instant and concise interpretation of essential
business information, such as the current transportation
time for a shipment, the current transportation costs, or the
utilization of the transportation vehicle.
Additionally, the S&R system detects situations early
which are relevant for the planning and coordination of
the logistics, such as delays of a freight or loading the
freight into a wrong container. In the case of such problem
situations, the S&R system commences arrangements in
order to deliver the goods in time, such as changing the
transportation route (e.g. choosing a more direct
transportation route to the customer) or changing the
method of transportation (e.g. express transportation
services). In case that it is not possible to deliver the
goods on time, the Sense and Respond system
automatically sends notification to the customers with an
estimate of the shipment delay.
In this example, the S&R system reacts in near real-
time to changes in the business environment. Events from
various sources (vehicles, distribution centers, contractors,
customers) are received and unified ( Event
Transformation) in order to assess the current state of the
business environment. Certain event patterns describe a
business situation (e.g. a truck is stuck in a traffic jam)
that is automatically discovered by the S&R system (
Situation Discovery). A business situation triggers the
invocation of analytical services in order to forecast
whether a shipment is going to be late ( Analytical
Processing). Based on the analytical results a rule decides
( Decision Making) whether the transportation route
should be adjusted, or whether the customers should be
notified about the shipment delay. The S&R system
instantaneously initiates and executes the appropriate
actions ( Response Management) based on the outcome
of the decision rule.
6. Conclusion and Future Work
Traditional BI architectures lack in the support of real-
time BI and closed-loop decision making. In this paper we
extended a traditional BI architecture with S&R system
and analytical services to transform business events into
performance indicators and intelligent business actions.
We discussed the latencies during the analytical
processing and showed how our extensions enable real-
time analytics for a business environment. The work
presented in this paper is part of a larger, long-term
research effort aiming to develop a service-oriented
Business Intelligence platform.
References
[1] Berry, M.J.A. and Linoff, G.S., Mastering Data
Mining: The Art and Science of Customer
Relationship Management John Wiley & Sons, Inc,
New York, Chichester et al., 1999.
[2] Bruckner, R. M., List, B. and Schiefer, J., Striving
Towards Near Real-Time Data Integration for Data
Warehouses, In Proc. of the 4th Intl. Conf. on Data
Warehousing and Knowledge Discovery (DaWaK
2002), Springer LNCS 2454, pp. 317–326, Aix-en-
Provence, France, Sept. 2002.
[3] Brobst, S. and Ballinger, C., Active Data
Warehousing. Whitepaper EB–1327, NCR
Corporation, 2000.
[4] Codd, E.F., Codd, S.B. and Salley, C.T.: Providing
OLAP (On-Line Analytical Processing) to User
Analysts: An IT Mandate, Arbor Software
Corporation, 1999.
[5] D’Aveni, R. M., Hypercompetition. New York:
The Free Press, 1994.
[6] Davenport, T.H., Process Innovation:
Reengineering Work through Information
Technology, Harvard Business School Press,
Boston, 1993.
[7] Denison, D.R., Towards a process-based theory of
organizational design: Can organizations be
designed around value chains and networks? Adv.
Strategic Management 14, 1997, pp. 1-44.
[8] Dijksterhuis, M.S., Van den Bosch, F.A. J. and
Volberda, H.W., Where do new organizational
forms come from? Management logics as a source
of coevolution, Organization Science 10(5) 1999,
pp. 569-582.
[9] Eckerson, W.W., The decision support sweet spot,
Journal of Data Warehousing, 3:2, Summer, 1998,
2-7 and Gray, P. and Watson, H.J., Decision
support in the data warehouse, Prentice Hall,
Upper Saddle River, N.J., 1998.
[10] Evans and Wurster, Blown to bits. Boston:
Harvard Business School Press, 2000.
[11] Geishecker, L., Manage Corporate Performance to
Outperform Competitors, Gartner Group, note
COM-18-3797, 2002.
[12] Grigoria, D., Casatib, F., Castellanosb, M., Dayalb,
U., Sayalb, M. and Shan, M.-C., Business Process
Intelligence, in: Computers in Industry 53, 2004,
pp. 321-343.
[13] Hackathorn, R., Current Practices in Active Data
Warehousing,http://www.dmreview.com/whitepaper/WID489.pd
f accessed 10 March 2005.
[14] Hammer, M. and Champy, J., Reengineering the
Corporation, Nicholas Brealey Publishing,
London, 1993.
[15] Imhoff, C., The Corporate Information Factory,
DM Review, December 1999,http://www.dmreview.com/editorial/dmreview,
accessed 29 March 2000.
[16] Inmon, W. H, Imhoff, C. and Sousa, R., Corporate
Information Factory, Second Edition, J.Wiley and
Sons, New York, 2001.
[17] Inmon, W.H. and Zachman, J.A., Geiger, J.G.,
Data Stores Data Warehousing and the Zachman
Framework, McGraw-Hill: New York et al. 1997.
[18] Inmon, W.H., Building the Operational Data Store,
2nd edition, Wiley: New York et al. 1999.
[19] Mahoney, J., The New Focus of IT Value:
Externalizing Agile Business, Gartner Research
Note, 17 July 2002.
[20] Malhotra, Y., From information management to
knowledge management: Beyond “Hi-Tech
Hidebound” systems, in Srikantaiah, T. K. and
Koenig, M.E.D. (Eds.) Knowledge Management,
Medford, NJ, 2000.
[21] Martin, W., Business Performance Management –
Efficiently Managing Business Processes. Research
Bulletin, 2003,http://www.it-research.net,
accessed 12 May 2003.
[22] Moncla, B. and Arents-Gregory, M., Corporate
Performance Management: Turning Strategy into
Action, DM Review, December, 2003,http://www.dmreview.com/editorial/dmreview,
accessed 15 December 2003.
[23] OVUM Evaluates, CRM Strategies: Technology
Choices for the Customer-focussed Business;
OVUM Ltd., London 1999.
[24] Raden, N., Exploring the Business Imperative of
Real-Time Analytics, Teradata white paper,
October 2003.
[25] Schiefer, J. and Bruckner, R. M., Container-
Managed ETL Applications for Integrating Data in
Near Real-Time, in: Proceedings of the
International Conference on Information Systems
(ICIS), pp. 604-616, 2003.
[26] Schulte, R., Application Integration Scenario: How
the War is Being Won, in: Gartner Group (Ed.):
Application Integration – Making E-Business
Work, London, 6-7 September 2000.
[27] Simon, H.A., The new science of management
decisions, Prentice Hall, Englewood Cliffs, N.J.,
1960.
[28] Sprague, R.H., A framework for the development
of decision support systems, MIS Quarterly, 4(4),
1980, 7-32; and Silver, M.S., Systems that support
decision-makers: Description and analysis. John
Wiley & Sons, New York, 1991.
[29] Tushman, M.L. and Nadler, D.A., Information
processing as an integrating concept in
organization design, Academy of Management
Review, 3, 1978, pp. 613-624.
[30] Weiner, J.C., Cybernetics, MIT Press, Cambridge,
Ma, 1948.
[31] Williams, S. and Williams, N., The Business Value
of Business Intelligence, in: Business Intelligence
Journal, Fall, 8, 4, 2004.
[32] Winter, R., The Current and Future Role of Data
Warehousing in Corporate Application
Architecture, in Proceedings of the 34th Hawaii
International Conference on System Sciences –
2001.
doc_700023529.pdf
In the 21st century, organizations are experiencing environmental changes characterized by indistinct organizational boundaries and fast-paced change.
Enhanced Business Intelligence - Supporting Business
Processes with Real-Time Business Analytics
Andreas Seufert
University of Applied Science Ludwigshafen
[email protected]
Josef Schiefer
Institute for Software Technology
and Interactive Systems
[email protected]
Abstract
In the 21st century, organizations are experiencing
environmental changes characterized by indistinct
organizational boundaries and fast-paced change. As a
result firms need appropriate decision support
infrastructures in order to face these challenges. Current
Data Warehousing and Business Intelligence approaches
are widely accepted as a middleware layer for state-of–
the-art decision support. However, they do not provide
sufficient support in dealing with the upcoming
challenges, such as real-time and closed loop decision
making. In this paper, we suggest an architecture for
enhanced Business Intelligence that aims to increase the
value of Business Intelligence by reducing action time
and interlinking business processes into decision making.
1. Introduction
In the 21st century, organizations are evolving into new
forms based on knowledge and networks in response to an
environment characterized by indistinct organizational
boundaries and fast-paced change. Firms are experiencing
environmental changes resulting from the new economics
of information [10] and the increasingly dynamic and
global nature of competition [5] . Therefore organizational
survival depends on the construction and integration of
knowledge fostering the adaptation to the environment, as
well as stimulating environmental changes through the
firm’s knowledge and practices [8]. As a result especially
investments in IT that enable differentiation are of ever
increasing importance [19].
An important component of this investment is in
Business Intelligence (BI). Coined by Gartner in the early
1990s, the term BI denotes on the one hand an analytic
process that transforms internal and external data into
information about capabilities, market positions, activities,
and goals that the company should pursue in order to stay
competitive. On the other hand, BI stands for Information
System concepts like Online Analytical processing
(OLAP), Querying and Reporting, or Data Mining that
provide different methods for a flexible goal-driven
analysis of business data, provided through a central data
pool.
BI may facilitate the connections in the new-form
organization, bringing real-time information to centralized
repositories and support analytics that can be exploited at
every horizontal and vertical level within and outside the
firm [20].
This research aims at contributing to the future
development of BI. It is grounded in the working
assumptions of the information processing theory [29]. As
BI is a newer form of decision support systems [9], [28]
we build on Simon’s [27] perspective of decision making
and Weiner’s classic model of an organization as an
adaptive system [30]. Therefore, we conceptualize inputs
being processed into outputs which feed back to influence
inputs and enable adaptation to external uncertainty.
The remainder of this paper is structured as follows:
Section 2 reviews related work and describes major future
challenges for decision support and Business Intelligence.
Section 3 introduces a framework for enhanced Business
Intelligence integrating the process-oriented and real-time
analytics perspective. In Section 4-5 we describe an
architecture for enhancing Business Intelligence and its
contribution for advanced decision support.
2. Related Work
Due to the challenges in the competitive environment,
two major challenges of the traditional business
intelligence (BI) concept can be identified.
Convergence of Business Processes and BI
First, the common thread running through the new-
form organization is that the organization’s foundation is
the value chain – a set of primary and secondary activities
that create value for customers [7]. Therefore, BI is
targeted to support process-oriented organizations.
Initiated by the works of Hammer and Champy [14],
Davenport [6] and others, companies have redesigned
their organizations around business processes. With
process orientation gaining importance, the need for
effectively managing and controlling business processes is
of great importance resulting in new requirements for
decision support. Collecting and reconciling all
operational data related to business processes, enables the
measurement of process performance and helps to identify
opportunities for process improvement.
As management decisions require the integration of
decision-relevant information, operational data from
different applications have to be collected, integrated and
prepared for data analysis. Today, this is mainly achieved
by using traditional data warehouse systems and
traditional business intelligence tools [31]. In most cases,
the resulting management information systems are not
primarily targeted on measuring the performance of
business processes but on fulfilling traditional reporting
requirements (e.g. financial reporting) [12].
Convergence of BI and EAI
Second, Business Intelligence aims at providing a
closed-loop support that interlinks strategy formulation,
process design and execution with business intelligence
[11], [22].
In order to increase their competitiveness, companies
strive towards reducing the time needed to react to
relevant business events. An ideal state would be reached
if reactions were possible in real-time, i.e. without any
latency between recognizing a relevant business event and
taking an appropriate action. Enterprise Application
Integration (EAI) suites provide a popular solution for
integrating heterogeneous applications in or near real-time
because they are able to seamlessly publish any kind of
data updates to every subscribing application.
However, this integration usually is achieved between
operational systems and involves only little data
consolidation. When it comes to extensive data analysis,
Business Intelligence will be used to produce the
information that is necessary to decide and take
appropriate actions. Addressing this field, real-time
decision support gained great attention. Concepts such as
active warehousing [3], real-time analytics [24], real-time
warehousing Error! Reference source not found., or
real-time decision support provide suggestions of how to
speed up the flow of information in order to achieve
competitive advantage.
Additionally, the vendors of BI and data warehousing
solutions tend to enhance their products by mechanisms
for real-time data integration and real-time analysis: this
leads to the convergence of EAI and BI solutions [21].
3. Framework for enhanced BI
BI as Decision Support Middleware Layer
Data warehouse systems are widely accepted as a new
middleware layer between transactional applications and
decision support applications, thereby decoupling systems
tailored to an efficient handling of business transactions
from systems tailored to an efficient support of business
decisions. The transformation of transaction data into
integrated, consistent input for decision support
applications by data warehousing consumes a certain
amount of time and creates non-volatile, aggregate
information. Many operational decisions (e.g. promotion
effectiveness, customer retention, key account information
[17], need actual yet integrated and subject-oriented data
in or near real-time [26].
Basic orientation: Transaction Subject
Time references: Actual Actual + historical
Access: Read-write Read-only
Aggregation level: Detailed Aggregated
Integration level: Isolated Integrated
Accessibility: Real-time Delayed
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Basic orientation: Transaction Subject
Time references: Actual Actual + historical
Access: Read-write Read-only
Aggregation level: Detailed Aggregated
Integration level: Isolated Integrated
Accessibility: Real-time Delayed
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Data managed by transactional systems
Data managed by operational data stores
Data managed by data warehouse
Legend:
Figure 1: Operational data stores vs. Data Warehouse
(adapted from Winter [32])
Therefore, the concept of operational data stores has
been introduced for operational decision support [18] and
for data-oriented application integration [15]. It becomes
evident that operational data stores can be positioned
between transactional applications and the data warehouse
[17], [18]. Data managed by operational data stores have
characteristics that differ from data managed by
operational applications, as well as from data managed by
the data warehouse (see figure 1).
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
Vertical
applications
Cross-
product
applications
Operational
data stores
Data staging
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Channel oriented
integration
Product oriented
Integration
Core data
Integration
Business
function
Business
process
Business
unit
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
(Channel oriented
integration)
Vertical
applications
(Product oriented
Integration)
Cross-
product
applications
(core data
Integration)
Opera-
tional
data
stores D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
D
a
t
a
s
t
a
g
i
n
g
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Horizontal
applications
Vertical
applications
Cross-
product
applications
Operational
data stores
Data staging
Data warehouse
Extraction, transformation, integration, correction
Selection, aggregation, supplementation
Decision support applications
Channel oriented
integration
Product oriented
Integration
Core data
Integration
Business
function
Business
process
Business
unit
Figure 2: Business Intelligence as Decision Support
Middleware Layer (adapted from Winter [32])
Since both the data warehouse and operational data
stores can be regarded as data-oriented integration
middleware, it was proposed that operational data stores
should be implemented as a part of the data warehouse, or
that the data warehouse should be directly utilized for
transactional services like customer relationship
management or e-commerce [23]. Since operational data
stores are fundamentally different from the data
warehouse due to real-time processing needs and read-
write access to data, these two middleware layers are
usually separated in the application architecture.
Therefore, Winter suggests the application architecture for
decision support [32] shown in Figure 2.
Enhancing BI with Real-Time Business Analytics
Data warehousing may contribute to an efficient
information supply between transactional applications and
decision support applications. However, information
‘backflows’ take their way indirectly (i.e. by actions of
decision makers) into the transactional systems and not
directly through the data warehouse (open-loop approach).
Operational data stores, in contrast allow an efficient
‘local’ closed-loop supported between vertical and
horizontal transactional applications.
What is decisive is a closed loop approach interlinking
operational and strategic decision making. Therefore, BI
has to be enhanced towards a closed loop real-time BI,
shortening the period of time between the occurrence of a
business event that requires an appropriate action by the
organization and the time the action is finally carried out.
According to Hackathorn [13], the additional business
value of an action decreases, the more time elapses from
the occurrence of the event to taking action.
d
a
ta
la
te
n
c
y
Figure 3: Business value and reduced action time
(adapted from Hackathorn [13])
The elapsed time is called action time and can be seen
as the latency of an action. Action time comprises four
components:
1. Data latency is the time from the occurrence of the
business event until the data is stored and ready for
analysis.
2. The time from the point when data is available for
analysis to the time when information is generated out
of it is called analysis latency; it includes the time to
determine root causes of business situations.
3. Decision latency is the time it takes from the delivery
of the information to selecting a strategy in order to
change the business environment. This type of latency
mostly depends on the time the decision makers need to
decide on the most appropriate actions for a response
to the business environment.
4. Response latency is the time needed to take an action
based on the decision made and to monitor its outcome.
This includes communicating the decision made as a
command or suggestion, or executing a business action
in a target system.
4. Architecture for Enhancing BI
In this section, we propose an architecture for real-time
analytics with the aim of reducing the action time and
thereby increasing the value of Business Intelligence.
Figure 4 shows an architectural diagram for real-time
analytics with two infrastructure types: 1) information
integration infrastructure and 2) business integration
infrastructure. The main objective of this architecture is to
seamlessly integrate the two infrastructure types in order
to minimize the aforementioned latencies. In the diagram
we extended the components and modules of traditional
BI in order to enable real-time analytics for business
environments.
The information integration infrastructure is
responsible for managing the data for business intelligence
purposes and offers data analysis to decision makers and
to IT systems. Traditional Business Intelligence aims to
support strategic decision makers and therefore uses
analytical applications that are periodically fed with data
from the data warehouse. These analytical applications are
generally completely disconnected from operational IT
systems. Decisions are executed by communicating them
as a command or suggestion to humans. On the other
hand, the enhanced Business Intelligence includes
analytical services which are continuously fed with data
from the operational environment (e.g. via the ODS) and
can be directly invoked by other systems. The object of
analytical services is to provide continuous data analysis
that is able to also cope with current changes in the
business environment.
The central piece of the Business Integration
infrastructure is a Sense & Respond (S&R) system that
communicates events via hubs with the internal and
external business environment. Table 1 provides an
overview of the modules of the S&R system. The internal
business environment comprises vertical and horizontal
applications which are shown on the left side of the
diagram. From the external business environment on the
right side, events are captured during the collaboration
with business partners, or when the contents of websites
changes (e.g. a competitor updates the product prices).
Information Integration Information Integration
Infrastructure Infrastructure
Analytical Services Analytical Services
Sense & Sense & Respond Respond
Horizontal Horizontal
Applications Applications
Horizontal Horizontal
Applications Applications
Business Integration Business Integration
Infrastructure Infrastructure
Analytical Processing Analytical Processing
Decision Making Decision Making
Response Management Response Management
Situation Detection Situation Detection
Event Transformation Event Transformation
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
Events
Metrics
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
H
u
b
H
u
b
H
u
b
H
u
b
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
Events
Metrics
Events
Metrics
Events/ Metrics
Situations
Situations
Actions
Actions
Events/ Metrics
Situations
OLAP / OLAP / Data Data Mining Mining
Data Data Warehouse Warehouse / ODS / ODS
Traditional Business Intelligence Enhanced Business Intelligence
Vertical Vertical
Applications Applications
Vertical Vertical
Applications Applications
Analytical Applications Analytical Applications
H
u
b
H
u
b
H
u
b
H
u
b
Web Web
Information Information
Collaboration with Collaboration with
Business Partners Business Partners
Figure 4: Architecture with enhanced Business Intelligence
Module Description
Event
Transformation
Transformation of the captured events from the
business environment into meaningful business
information, such as key performance indicators.
Situation Discovery Detection of business situations and exceptions in
the event data and business information. For
instance, an organization wants to detect
suspicious customer behavior (e.g. fraud behavior)
which should be countered with a proactive
response.
Analytical
Processing
Invocation of analytical services in order to
determine the root causes for business situations
and exceptions. Also, prediction of the
performance and assessment of the risks for
changing the business environment. Please note,
that the S&R system delegates the analytical
processing to the analytical services of the
information integration infrastructure. The S&R
system correlates and collects event data for the
data analysis, invokes the analytical services and
processes the results.
Decision Making Based on the analytical results, selection of the
best option for improving the current business
situations and determining the most appropriate
action for a response to the business environment.
This step can be automated with rules, or done by
involving humans.
Response
Management
Response to business environment by
communicating the decision made as a command
or suggestion (e.g. by e-mail), or by directly
adapting and reconfiguring business processes and
IT systems.
Table 1. Modules of the S&R System
During the event processing in the S&R system,
business information is continuously generated and
decisions are made to which a response follows. The
response has an effect on the source systems (from which
the S&R system originally received the events) and
consequently also on the performance and the success of
the organization. In order to reduce action time, latency
has to be reduced in all four stages of the action time. In
the following we discuss how our extensions to traditional
Business Intelligence help to minimize the various types
of latencies.
Minimizing Data Latency
Data Latency is usually mainly influenced by the
refresh cycle of the data warehouse system in which the
data is stored for analysis purposes. A main drawback of
the data warehouse concept is the time-consuming and
resource-intensive process of extracting data from
operational systems, transforming it and loading it into the
data warehouse database. Due to this fact, the so-called
ETL processing is often executed in batch mode at non-
peak times (e.g. overnight), causing time-lags between the
recognition of a business event and its delivery for
analysis purposes.
In contrast to the data warehouse system that requires
extensive data cleansing, data consolidation and data
quality management, an ODS stores a limited scope of
data with only basic (or even none) consolidation and data
quality management, thereby allowing real-time or near
real-time updates and faster data distribution [16].
Another way of reducing data latency is to change from
a periodic batch-oriented to an event-driven update of the
data warehouse by using EAI technology. By doing this,
data representing a certain business event will be
populated into the data warehouse as soon as the event is
recognized in an operational system [25].
We extend the approach from Schiefer and Bruckner
[25] by integrating event streams from the business
environment with integration hubs. The integration hubs
unify the raw event data and feed them continuously into
the S&R system. The S&R system generates business
metrics that are stored in a real-time data store (e.g. an
ODS) and used as input for the invocation of analytical
services.
Minimizing Analysis Latency
Analysis latency is mainly determined by the time it
takes to inform the person in charge of data analysis that
new data has to be analyzed, the time needed to choose
appropriate analysis models and the time to process the
data and present the results. Current approaches that deal
with the reduction of analysis latency are provided by BI
software. A prominent BI concept is Online Analytical
processing (OLAP), which concentrates on reducing the
time needed for analyzing the data by providing powerful
user-interfaces that let the analyst explore the data along
previously defined analysis dimensions [4].
In contrast to the fixed dimensional structures
necessary for doing OLAP, Data Mining is a more flexible
BI approach which allows the application of different data
exploration techniques to a large amount of data in order
to discover unknown relationships between variables or
single data items [1]. While pure data mining is mainly
focused on reducing the time for data processing by
applying efficient data exploration algorithms, data
mining software tools like SPSS or IBM intelligent miner
also facilitate the process of selecting and adapting a data
mining technique that is suitable for a given problem. In
order to reduce the time for notifying the analyst of
business events, the approach of automated exception
reporting can be used.
In order to reduce the analysis latency, we use in our
architecture analytical services in the information
integration infrastructure which enable an invocation of
analytical function without manual intervention. A
service-oriented architecture for analysis provides
standardization for using various types of analysis
techniques within a BI process. Unlike traditional BI
tools, analysis services are permanently available, which
is a prerequisite for a continuous and automated BI
process.
Minimizing Decision Latency
The least IT support can be identified in the area of
decision latency. In most cases, the interpretation of
analysis results and the derivation of appropriate actions
are seen as manual processes that have to be carried out
by knowledge workers and are therefore time consuming.
Newest advances especially in the area of Business
Activity Monitoring (BAM) try to improve this situation
by automating certain decision processes with the help of
rule-based decision engines. Based on the real-time
analysis of data from an EAI platform, the decision engine
checks for predefined business rules and notifies
responsible people, or triggers other tools for conducting
further actions.
In our architecture we follow the approach from BAM
and use rule-based decision making for automating many
operational and tactical decisions. Please note, that the
rules for these decisions are in many cases derived from
strategic decisions that are made by humans. The decision
rules help to intelligently respond very quickly to the
current business situation.
Minimizing Response Latency
The final outcomes of a Business Intelligence process
are actions based on the decisions made (manually or
automated). There are two kinds of response latencies: 1)
the time it takes to initiate an action and 2) the time it
takes to execute and monitor the action. Hackathorn
considers in [13] the response latency as part of the
decision latency which is reasonable for strategic decision
making. When it comes to real-time analytics, it is also
crucial to communicate and carry out a decision very
quickly. We consider separately in our model the time it
takes to initiate and execute an action in the business
environment. For this reason, a S&R system includes a
response management module which is responsible for
triggering business operations and monitoring their
outcome.
The business value of our architecture with enhanced
Business Intelligence can be summarized as follows:
- Real-time business information: Minimal latency for
preparing and analyzing data and hence improved
visibility and accuracy of business performance
indicators. The indicators are also available for
operational and tactical decisions.
- Optimized business processes: By integrating real-
time analytics, the internal and external business
environment can be optimized by more efficient and
intelligent control mechanisms. The S&R system
operates as an “external advisor” for the business
environment.
- Automatic discovery of situations and exceptions:
A S&R system supports a continuous discovery of
business opportunities and exceptions based on the
current state of the business environment. For instance,
companies are able to detect suspicious customer
behavior (e.g. fraud behavior) which can be countered
with a proactive response.
- Proactive responses: By continuously observing and
analyzing customers, business partners and the
competition, the business environment can be
proactively adapted and optimized.
- Generating more accurate forecasts in near real-
time under consideration of current and historic data
(e.g. continuous update and optimization of production
plans based on the current orders).
- Integrating internal and external source systems:
The S&R system is able to correlate and merge event
streams from the internal and external business
environment.
- Less integration effort: The S&R system has a
significantly lower integration effort than traditional
data warehouse solutions since only events from the
source system have to be integrated. For the event
processing the S&R system does not have to know
internal details of operational systems, such as the data
model.
5. Real-Time Business Analytics in Action
Let’s assume that the company Meyer AG offers
transportation services for various goods. In order to
better monitor, synchronize and optimize the processing
flows, the company uses mobile devices (e.g. mobile bar
code scanners) to store information for every article about
the current location and the status of the current method of
transportation. The S&R system continuously processes
and analyses this data and calculates indicators which
provide instant and concise interpretation of essential
business information, such as the current transportation
time for a shipment, the current transportation costs, or the
utilization of the transportation vehicle.
Additionally, the S&R system detects situations early
which are relevant for the planning and coordination of
the logistics, such as delays of a freight or loading the
freight into a wrong container. In the case of such problem
situations, the S&R system commences arrangements in
order to deliver the goods in time, such as changing the
transportation route (e.g. choosing a more direct
transportation route to the customer) or changing the
method of transportation (e.g. express transportation
services). In case that it is not possible to deliver the
goods on time, the Sense and Respond system
automatically sends notification to the customers with an
estimate of the shipment delay.
In this example, the S&R system reacts in near real-
time to changes in the business environment. Events from
various sources (vehicles, distribution centers, contractors,
customers) are received and unified ( Event
Transformation) in order to assess the current state of the
business environment. Certain event patterns describe a
business situation (e.g. a truck is stuck in a traffic jam)
that is automatically discovered by the S&R system (
Situation Discovery). A business situation triggers the
invocation of analytical services in order to forecast
whether a shipment is going to be late ( Analytical
Processing). Based on the analytical results a rule decides
( Decision Making) whether the transportation route
should be adjusted, or whether the customers should be
notified about the shipment delay. The S&R system
instantaneously initiates and executes the appropriate
actions ( Response Management) based on the outcome
of the decision rule.
6. Conclusion and Future Work
Traditional BI architectures lack in the support of real-
time BI and closed-loop decision making. In this paper we
extended a traditional BI architecture with S&R system
and analytical services to transform business events into
performance indicators and intelligent business actions.
We discussed the latencies during the analytical
processing and showed how our extensions enable real-
time analytics for a business environment. The work
presented in this paper is part of a larger, long-term
research effort aiming to develop a service-oriented
Business Intelligence platform.
References
[1] Berry, M.J.A. and Linoff, G.S., Mastering Data
Mining: The Art and Science of Customer
Relationship Management John Wiley & Sons, Inc,
New York, Chichester et al., 1999.
[2] Bruckner, R. M., List, B. and Schiefer, J., Striving
Towards Near Real-Time Data Integration for Data
Warehouses, In Proc. of the 4th Intl. Conf. on Data
Warehousing and Knowledge Discovery (DaWaK
2002), Springer LNCS 2454, pp. 317–326, Aix-en-
Provence, France, Sept. 2002.
[3] Brobst, S. and Ballinger, C., Active Data
Warehousing. Whitepaper EB–1327, NCR
Corporation, 2000.
[4] Codd, E.F., Codd, S.B. and Salley, C.T.: Providing
OLAP (On-Line Analytical Processing) to User
Analysts: An IT Mandate, Arbor Software
Corporation, 1999.
[5] D’Aveni, R. M., Hypercompetition. New York:
The Free Press, 1994.
[6] Davenport, T.H., Process Innovation:
Reengineering Work through Information
Technology, Harvard Business School Press,
Boston, 1993.
[7] Denison, D.R., Towards a process-based theory of
organizational design: Can organizations be
designed around value chains and networks? Adv.
Strategic Management 14, 1997, pp. 1-44.
[8] Dijksterhuis, M.S., Van den Bosch, F.A. J. and
Volberda, H.W., Where do new organizational
forms come from? Management logics as a source
of coevolution, Organization Science 10(5) 1999,
pp. 569-582.
[9] Eckerson, W.W., The decision support sweet spot,
Journal of Data Warehousing, 3:2, Summer, 1998,
2-7 and Gray, P. and Watson, H.J., Decision
support in the data warehouse, Prentice Hall,
Upper Saddle River, N.J., 1998.
[10] Evans and Wurster, Blown to bits. Boston:
Harvard Business School Press, 2000.
[11] Geishecker, L., Manage Corporate Performance to
Outperform Competitors, Gartner Group, note
COM-18-3797, 2002.
[12] Grigoria, D., Casatib, F., Castellanosb, M., Dayalb,
U., Sayalb, M. and Shan, M.-C., Business Process
Intelligence, in: Computers in Industry 53, 2004,
pp. 321-343.
[13] Hackathorn, R., Current Practices in Active Data
Warehousing,http://www.dmreview.com/whitepaper/WID489.pd
f accessed 10 March 2005.
[14] Hammer, M. and Champy, J., Reengineering the
Corporation, Nicholas Brealey Publishing,
London, 1993.
[15] Imhoff, C., The Corporate Information Factory,
DM Review, December 1999,http://www.dmreview.com/editorial/dmreview,
accessed 29 March 2000.
[16] Inmon, W. H, Imhoff, C. and Sousa, R., Corporate
Information Factory, Second Edition, J.Wiley and
Sons, New York, 2001.
[17] Inmon, W.H. and Zachman, J.A., Geiger, J.G.,
Data Stores Data Warehousing and the Zachman
Framework, McGraw-Hill: New York et al. 1997.
[18] Inmon, W.H., Building the Operational Data Store,
2nd edition, Wiley: New York et al. 1999.
[19] Mahoney, J., The New Focus of IT Value:
Externalizing Agile Business, Gartner Research
Note, 17 July 2002.
[20] Malhotra, Y., From information management to
knowledge management: Beyond “Hi-Tech
Hidebound” systems, in Srikantaiah, T. K. and
Koenig, M.E.D. (Eds.) Knowledge Management,
Medford, NJ, 2000.
[21] Martin, W., Business Performance Management –
Efficiently Managing Business Processes. Research
Bulletin, 2003,http://www.it-research.net,
accessed 12 May 2003.
[22] Moncla, B. and Arents-Gregory, M., Corporate
Performance Management: Turning Strategy into
Action, DM Review, December, 2003,http://www.dmreview.com/editorial/dmreview,
accessed 15 December 2003.
[23] OVUM Evaluates, CRM Strategies: Technology
Choices for the Customer-focussed Business;
OVUM Ltd., London 1999.
[24] Raden, N., Exploring the Business Imperative of
Real-Time Analytics, Teradata white paper,
October 2003.
[25] Schiefer, J. and Bruckner, R. M., Container-
Managed ETL Applications for Integrating Data in
Near Real-Time, in: Proceedings of the
International Conference on Information Systems
(ICIS), pp. 604-616, 2003.
[26] Schulte, R., Application Integration Scenario: How
the War is Being Won, in: Gartner Group (Ed.):
Application Integration – Making E-Business
Work, London, 6-7 September 2000.
[27] Simon, H.A., The new science of management
decisions, Prentice Hall, Englewood Cliffs, N.J.,
1960.
[28] Sprague, R.H., A framework for the development
of decision support systems, MIS Quarterly, 4(4),
1980, 7-32; and Silver, M.S., Systems that support
decision-makers: Description and analysis. John
Wiley & Sons, New York, 1991.
[29] Tushman, M.L. and Nadler, D.A., Information
processing as an integrating concept in
organization design, Academy of Management
Review, 3, 1978, pp. 613-624.
[30] Weiner, J.C., Cybernetics, MIT Press, Cambridge,
Ma, 1948.
[31] Williams, S. and Williams, N., The Business Value
of Business Intelligence, in: Business Intelligence
Journal, Fall, 8, 4, 2004.
[32] Winter, R., The Current and Future Role of Data
Warehousing in Corporate Application
Architecture, in Proceedings of the 34th Hawaii
International Conference on System Sciences –
2001.
doc_700023529.pdf