Description
In the last decade Business Intelligence has become one of the hottest areas in Information Technology.
Issues in Informing Science and Information Technology Volume 8, 2011
Would Cloud Computing Revolutionize Teaching
Business Intelligence Courses?
Stevan Mrdalj
Eastern Michigan University, Ypsilanti, MI, USA
[email protected]
Abstract
In the last decade Business Intelligence has become one of the hottest areas in Information Tech-
nology. Meanwhile, teaching Business Intelligence courses for Masters of Business Administra-
tion programs faces a variety of challenges. One of the major hurdles in offering Business Intelli-
gence courses is the complex and costly computing infrastructure necessary for the software used
in such courses. One possible remedy might be to use cloud computing. Cloud computing offers
flexible access the necessary software and hardware through the Internet. It is becoming increas-
ingly popular because of its on-demand and pay-per-use business models. In this paper, we exam-
ine how cloud computing might be used to revolutionize the way of teaching business intelligence
courses. In doing so, we present various possible infrastructure scenarios and the necessary tools.
Finally, we discuss the perceived difficulties in adopting Business Intelligence as a service busi-
ness model in academic settings. We also reflect on the economics of cloud computing.
Keywords: Cloud Computing, Business Intelligence, Software as a Service, Teaching BI, Pay-
per-use.
Introduction
For the last several years, Masters of Business Administration (MBA) programs have been under
mounting pressure to provide courses to prepare their students to become the next generation of
Business Intelligence (BI) workers. In a recent survey of professors from more than 80 universi-
ties around the world (Betts & Kanaracus, 2010), Barbara Wixom discovered that 43% of the re-
spondents reported that they can’t adequately teach BI courses” because they lack access to the
needed software, hardware and real-world business problems”. She also indicates that “It’s clear
from the survey that instructors want to change the way students learn about business intelli-
gence. Professors want to provide large data sets, contemporary software tools, and real-world
content within their classrooms. But, factors like high technology costs, complex maintenance
requirements, and steep learning curves present insurmountable obstacles.” With the present eco-
nomic crisis and its repercussions on the academic environments, one of the major hurdles in of-
fering Business Intelligence courses is
the complex and costly computing infra-
structure necessary for the software used
in such courses. On the other hand,
unless Business Intelligence becomes a
mainstream field at business schools, the
shortage of business intelligence related
professions will continue to rise.
Material published as part of this publication, either on-line or
in print, is copyrighted by the Informing Science Institute.
Permission to make digital or paper copy of part or all of these
works for personal or classroom use is granted without fee
provided that the copies are not made or distributed for profit
or commercial advantage AND that copies 1) bear this notice
in full and 2) give the full citation on the first page. It is per-
missible to abstract these works so long as credit is given. To
copy in all other cases or to republish or to post on a server or
The cloud computing, also known as
software-as-a-service, delivery model
has the potential to become a major
to redistribute to lists requires specific permission and payment
of a fee. Contact [email protected] to request
redistribution permission.
Would Cloud Computing Revolutionize BI Courses
force in the future of offering BI courses. After a decade of growth and adoption, the commercial
cloud computing offerings came into existence and products like Amazon EC2, Microsoft Azure
and Salesforce.com helped to popularize it. The concept of on-demand BI promises to make it
affordable for business schools to deliver quality BI instruction to MBA students.
Computer Science programs were among the first to develop experimental undergraduate data
mining courses as reported by Lu and Bettine (2003). Banks, Dong, Liu and Mandvikar (2004)
report on teaching data mining courses as “an exciting addition to the curriculum at the sen-
ior/graduate level” that provide the opportunity to apply computer related education to various
domains and applications. Paper by Roiger (2005) presents a very good tutorial on data mining
that can be used by educators as an introduction to business intelligence. The typical structure of
a data warehousing and data mining course in the Computer Science curriculum is well described
by Fang and Tuladhar (2006) and Musicant (2006). Information systems programs at business
schools typically offer BI courses as elective courses (Watson, 2006). A good example of an ap-
plied BI course that is designed to appeal to the graduate students of the CIS programs as well as
MBA students selecting the CIS concentration is presented by Mrdalj (2007). Specifically, the
structure and components of this course focus on “business aspects” of data mining and data
warehousing by having students learn how to answer today’s BI questions. In one of the rare pa-
pers related to MBA programs, Mrdalj and Diallo (2010) present the structure and components of
a BI module that is designed to appeal to students of the finance program as well as to MBA stu-
dents selecting the finance concentration.
In the next section of this paper we stage our proposed solution with a brief presentation of the
typical computing infrastructure architecture needed for teaching BI courses with a discussion of
possible on-premises installations. It is followed by the description of cloud computing and its
possibilities in teaching BI courses. We conclude with perceived difficulties in using cloud com-
puting as a platform for teaching BI courses.
Computing Requirements for BI Courses
The typical computing infrastructure architecture needed for teaching BI courses is shown in Fig-
ure 1. It requires a database server to host source databases as well as the data warehouse.
Data Warehousing
OLAP
Cubes
BI Tools
Data
Sources
INSTRUCTOR
STUDENT
ERP/
CRM
OLTP
Legacy
System
Extract
Clean
Transform
Load
Figure 1: A Typical BI Architecture
210
Mrdalj
Next, we need data warehousing development tools to extract data from the sources and to clean,
transform and load data into the data warehouse. Although this part is not visible to the MBA stu-
dents, it is required for the instructors in order to provide the students with "real-life" data sets so
they can make sure that students learn from actual company data instead of textbook examples.
In addition, BI architecture requires technology such as Online Analytical Processing (OLAP)
that provides a platform for end-user based utilization of data warehouse (Larson, 2009) and is a
dominant paradigm in BI. This technology is essential for MBA students who are usually not fa-
miliar with SQL. OLAP tools give students flexible and fast access to multi-dimensional views
(such as store, account, time, geography) and to measures (such as sales, payments, discounts),
both summarized and detailed. Such access if further supported with standard OLAP operations
that include drill-down, roll-up, dice and slice. Another benefit of using OLAP Cubes is that stu-
dents can access them with tools such as the Microsoft Excel PivotTable (Harts, 2008; Mrdalj &
Diallo, 2010) as illustrated in Figure 2.
Tools like the Microsoft SQL Server Data Mining Add-ins (MacLennan, Tang & Crivat, 2009)
and SAS
®
Enterprise Miner (Cerrito, 2006) enable Microsoft Excel to become the analytical cli-
ent to the OLAP analytical servers. This enables MBA students to access multi-dimensional data
sources and to perform reporting, ad-hoc analysis and data mining within the very familiar envi-
ronment of Microsoft Excel (Mrdalj & Diallo, 2010). Consequently, Microsoft Excel, perhaps the
oldest BI tool, has regained its role in MBA education as an affordable BI front end tool.
The above described BI architecture can be implemented on-premises in several different ways.
We will discuss the following three most common approaches.
Stand Alone Infrastructure
The simplest and most restrictive is the stand-alone approach when the entire BI architecture (da-
tabase server, analytical server and BI tools) is installed on each individual workstation in the lab
or computer classroom. It requires early setup of all the software and data sets which can be eas-
ily cloned to all workstations. To accommodate the continuity in students’ work, such setup
would require that students use the very same workstation throughout the entire semester and to
keep their own backups. Each change in software and data sets during a semester would require a
manual installation of such changes on all workstations since re-cloning would diminish the pre-
vious student’s work. This setup would also require that students would need to do their assign-
OLAP
Figure 2: End-user BI Access
211
Would Cloud Computing Revolutionize BI Courses
ments and projects on very same workstations that they use during classes or labs. Such a re-
quirement poses serious constraints on the type and quantity of assignments that can be given to
students. Otherwise, students would need to have the complete BI architecture installed on their
laptops/desktops.
Setting up a proper BI environment on students’ laptops is a multi-step process (for illustration
purposes we will describe a scenario with Microsoft SQL Server). First, it requires installing the
SQL Server. The next step is to install sample data warehouses including the corresponding
OLAP Cubes. Unfortunately, deployment of OLAP Cubes requires some familiarity with the Mi-
crosoft SQL Server Management Studio or the Microsoft Business Intelligence Development
Studio project. The next step is to download and install Microsoft SQL Server Data Mining Add-
ins. The last task of this process is to connect Excel to the SQL Server Analysis Service and to
configure an analysis database to be used by Excel to perform data mining tasks. The above de-
scribed multi-step installation process would certainly provide a major challenge to MBA stu-
dents as well as the necessity to acquire all the needed software. Based on our experience in using
this approach, we consider this to be prohibitive for the majority of MBA students and, therefore,
we consider it a most restrictive approach for teaching BI courses.
Local Area Network Infrastructure
The step-up approach would have a local area network where the database server hosting the data
warehouse and the analysis server hosting the OLAP cubes are installed on the network server
and all the client workstations would need only Excel with BI tool add-ins as shown in Figure 3.
Figure 3: Network Configuration
This approach certainly allows any mid-semester changes in the software and data sets. It requires
moderate DB administration since all clients would need proper authentication to access OLAP
cubes. The major problem with this approach is that those students wanting to do their assignment
on their laptops would need to be permanently connected to the network since they would need
access to the analysis server to perform BI tasks. Students who would like to do their work out-
side of the lab environment would need to have the stand-alone installation described above.
Web Enabled Infrastructure
To overcome the problems with the previous two approaches, business schools would need to
invest in the web enabled solution depicted in Figure 4. The advantage of this approach is that
students would only need to install Excel add-ins and to have available Internet access. On the
other hand, business schools would need considerable hardware and software resources to im-
212
Mrdalj
plement it. In addition, this approach would require a high level of expertise to configure database
and web servers as well as to make them operational non-stop. The cost of designing, building,
installing and running the web enabled infrastructure continues to be the most noticeable limiting
factor for business schools. Frequently such prohibitive costs have been the main limitation for
proliferation of BI courses at MBA programs.
Figure 4: Web Enabled Configuration
Another variation of this approach is to use remote access tools for the remote clients. Such an
approach would even eliminate the need for students to have Excel as well as BI tools installed on
their laptops. A comprehensive discussion of the remote access tools (Scher, 2010) is beyond the
scope of the paper, but we list some of them as examples: GoToMyPC, LogMeIn Pro2, Microsoft
Remote Desktop Connection, Remote PC and pcAnywhere.
If the universities do not have adequate resources or if they do not want to acquire and manage
technical resources or if they lack the necessary technical experience and skills, then they should
consider using cloud computing.
Using Cloud Computing for Teaching Business
Intelligence
Cloud computing is one of the most attractive technology areas today due to its cost efficiency
and flexibility. There are many definitions and interpretations of cloud computing, but for the
purpose of this paper we will use the following definition created by the National Institute of
Standards (Mell & Grance, 2009):
“Cloud computing is a model for enabling convenient, on-demand network access to a shared
pool of configurable computing resources (e.g., networks, servers, storage, applications, and ser-
vices) that can be rapidly provisioned and released with minimal management effort or service
provider interaction.”
In the cloud computing paradigm, the basic architecture needed for teaching BI courses based on
the three layered NIST service model (Mell & Grance, 2009) and its customizations for BI
(Chandra & Iyer, 2010; Reyes, 2010) is presented in Figure 5. At the lowest level, the processing
power refers to the hardware needed for running database servers and networking. Normally, the
customer does not control or manage this layer of the cloud infrastructure. The cloud’s Platform
as a Service (PaaS) provides instructors with the capability to deploy large data warehouses from
real businesses. The Software as a Service (SaaS) layer of the cloud provides students and in-
structors with applications running on a cloud infrastructure. These layers may be considered as
services to the layer above them.
213
Would Cloud Computing Revolutionize BI Courses
In relationship to the BI courses, the data warehousing tools would allow instructors to deploy
data warehouses and OLAP cubes as well as to maintain them over the course of the semester.
Students would gain access to the OLAP cubes and reporting, data mining, and business perform-
ance management applications. The OLAP cubes and applications are accessible from remote
clients’ devices trough a thin client interface such as a web browser (Mell & Grance, 2009) or
Excel add-ins (Predixion, 2010). In short, cloud computing shifts the physical construction of the
data center, creating network connections between students and the center, and providing all the
necessary software to the trusted service provider (Thomson, 2009).
Figure 5: Basic BI Architecture in the Cloud
Thanks to recent advances in technology in recent years, the commercial cloud computing offer-
ings may enable business schools to implement BI courses in a fraction of the time and avoid the
capital expenditure required by traditional installations (Thomson, 2009). Frequently mentioned
advantages of cloud computing are lower cost and non-stop operation. Even more important fea-
tures for the academic environments are summarized below (adapted from Katzan, 2009):
• The responsibility for hardware, application software, storage facilities, and professional
services resides with the provider.
• Systems software is available from a trusted vendor for supporting cloud services.
• Data centers are available for sustainable and reliable operation.
In other words, cloud computing would provide virtual and scalable infrastructure as a service
where the complexity of resource management is hidden from the client while enabling the client
to exploit supercomputing power on-demand without investing in huge infrastructure and man-
agement costs (Ananthanarayanan et al., 2009). A broad discussion of the available Cloud BI
tools is beyond the scope of this paper and it can be found in (Rayes, 2010; Tsai, 2009). Instead,
we list some of them as examples: Predixion Insight, CloudOLAP, IBM Smart Analytics Cloud,
Informatica 9, and Rockspace Cloud.
This brings the “thin client” architecture back to life where both instructors and students would
utilize the needed services from the cloud provided as shown in Figure 6. The biggest advantage
that cloud computing brings to instructors is that they can obtain the proof-of-concept for the de-
velopment of their courses in a matter of days. Instructors would be able to upload their example
214
Mrdalj
data warehouses and OLAP Cubes and make them available to students wherever Internet access
is available. Students would gain easy and location independent access without the need to ac-
quire complex and expensive software that needs to be installed on their workstations in order to
do their assignments.
OLAP
DW
ETL
STUDENT
INSTRUCTOR
Figure 6: Cloud BI
Perceived Difficulties in Using Cloud Computing
Despite many advantages and benefits that cloud computing brings to business schools there are
several perceived difficulties as well. In this paper, we will only concentrate on those that are
relevant to using cloud computing for teaching purposes. Many other challenges related to cloud
computing used for commercial purposes such as scalability, security, large data volume, speed of
access, and reliability can be found in (Reyes, 2010; Thomson, 2009).
Customization is one of the very important aspects of cloud computing used for teaching BI
courses. With non-configurable SaaS (Katzan, 2009), the cloud provider delivers a unique set of
applications to clients and limits instructors to tailor them to specific classwork needs. A similar
concern is for instructors teaching multiple sections when a multi-tenant SaaS would be required
to host common applications and unique data sets for different sections.
Another important consideration of adapting cloud computing for teaching BI courses is that its
pricing structure is variable and complex. There are three categories that we might observe as
valid options for the academic environments (adapted from Katzan, 2009):
• Perpetual license refers to an “up front” payment for service, and unlimited access for an
unlimited time.
• Subscription, as a form of cloud service monetization, can be conceptualized as a time-
based perpetual license, often applied to multiple users.
• Transaction based or variable pricing model, also known as pay-as-you-go plan, is based
on the data transfer volume and number of transactions.
The variable pricing model may look quite attractive for business schools since data volume and
number of transactions may not be large, but it would bring an uncertain cost associated with a
given course. Many would argue that the subscription model is more economical and that it pro-
215
Would Cloud Computing Revolutionize BI Courses
vides budgetary predictability and a lower cost over a longer period of time (Thomson, 2009).
What may drive up the cost for either pricing model are configurability and isolation. From a
pricing stand point non-configurable and single-tenant SaaS have a lower cost, but are quite pro-
hibitive in customization.
Both pricing models are something that business schools may not be accustomed to have associ-
ated with offering courses. Traditionally, universities have budgets for computer refresh programs
and laboratory maintenance and not for cloud subscription fees associated with an individual
course. Although either of the pricing models mentioned above might cost much less compared to
the costs of on-premises BI architectures, it would require a paradigm shift in university budget-
ing models.
Conclusions
We examined the requirements for implementing the BI architecture necessary for teaching BI
courses. Based on them, we concluded that cloud computing is an attractive solution for business
schools wanting to implement cost-effective, rapid and dynamic environments for their BI
courses. The promise of cloud computing is already proven at many companies and it might be
prudent for business schools to adopt this new approach instead of investing in ever more de-
manding and expensive computer labs and classrooms.
By outsourcing the physical construction of the data center, creating network connections be-
tween students and the center, and providing all the necessary software to students, business
schools can concentrate on rapid development of BI courses and their integration into MBA pro-
grams. With cloud computing, business schools may significantly reduce their costs while gaining
guaranteed performance and high student satisfaction.
We also hope that cloud-based BI service providers like Predixion, CloudOLAP and especially
Microsoft Azure (with the offering of the analysis server in the near future) would be willing to
provide academic discounts for their services to those universities teaching BI courses.
References
Ananthanarayanan, R., Gupta, K., Pandey, P., Pucha, H., Sarkar, H., Shah, M., & Tewari, R. (2009). Cloud
analytics: Do we really need to reinvent the storage stack? Proceedings of the 2009 Conference on Hot
Topics in Cloud Computing. Retrieved September 12, 2010 fromhttp://www.usenix.org/event/hotcloud09/tech/full_papers/ananthanarayanan.pdf
Banks, D. L., Dong, G., Liu, H., & Mandvikar, A. (2004). Teaching undergraduates data mining engineer-
ing programs. Proceedings of the 34th ASEE/IEEE Frontiers in Education Conference, October 20 –
23, 2004, Savannah, GA, T1A-1 - T1A-6. Retrieved September 22, 2010 fromhttp://www.public.asu.edu/~huanliu/papers/fie04.pdf
Betts, M., & Kanaracus, C., (2010, April 5). Professors cite challenges in teaching BI in college. Com-
puterworld. Retrieved November 10, 2010 fromhttp://www.computerworld.com/s/article/347880/Professors_Struggling_to_Teach_BI
Chandra, B., & Iyer, M. (2010). BI in a cloud: Defining the architecture for quick wins. SETLab Briefings,
8(1), 39-44. Retrieved October 10, 2010 fromhttp://www.infosys.com/research/publications/Documents/BI-in-a-cloud.pdf
Cerrito, P. B. (2006). Introduction to data mining using SAS Enterprise Miner. Cary, NC: SAS Institute.
Fang, R., & Tuladhar, S. (2006). Teaching data warehousing and data mining in a graduate program of in-
formation technology. Journal of Computing Sciences in Colleges, 21(5), 137-144.
Harts, D. (2008). Microsoft Office 2007 business intelligence: Reporting, analysis, and measurement from
the desktop. McGraw-Hill/Osborne.
216
Mrdalj
217
Katzan, H. (2009). Cloud software service: Concepts, technology, economics. Service Science, 1(4), 256-
269.
Larson, B. (2009). Delivering business intelligence with Microsoft SQL Server 2008. McGraw-
Hill/Osborne.
Lu, Y., & Bettine, J, (2003). Data mining: An experimental undergraduate course. Journal of Computing
Sciences in Colleges, 18(3), 81-86.
MacLennan, J., Tang, Z. H. & Crivat, B. (2009). Data mining with SQL Server 2008. Wiley Publishing.
Mell, P., & Grance, T. (2009). The NIST definition of cloud computing. National Institute of Standards and
Technology, Version 15, 10-7-09. Retrieved October 10, 2010 fromhttp://csrc.nist.gov/groups/SNS/cloud-computing/
Mrdalj, S (2007). Teaching an applied business intelligence course. Issues in Information Systems, 8(1),
134-138.
Mrdalj, S., & Diallo, A., (2010). Bringing business intelligence into finance curriculum. Issues in Informa-
tion Systems, 11(2), 168-175.
Musicant, D. R. (2006). A data mining course for computer science. ACM SIGCSE Bulletin, 38(1), 538-
542.
Predixion Insight. (2010). Predixion software. Retrieved November 15, 2010 fromhttps://www.predixionsoftware.com/predixion/Portals/0/pdf/Predixion_Insight_DS_FINAL_9_4_10.p
df
Reyes, E.P. (2010). A systems thinking approach to business intelligence solutions based on cloud comput-
ing. Boston, MA: Massachusetts Institute of Technology. Retrieved November 2, 2010 fromhttp://hdl.handle.net/1721.1/59267
Roiger, R. J. (2005). Teaching an introductory course in data mining. ACM SIGCSE Bulletin, 37(3), 415-
421.
Scher, R., (2010, October). Remote access tools. PC Today, 20-23.
Thompson, J. (2009). Business intelligence in a SaaS environment. Business Intelligence Journal, 14(4),
50-55. Retrieved October 5, 2010 fromhttp://tdwi.org/research/2009/12/business-intelligence-journal-
vol-14-no-4.aspx?tc=page0
Tsai, J. (2009, August). Intelligence in the cloud. Customer Relationship Management Magazine, 38-43,
Retrieved October 15, 2010 fromhttp://www.destinationcrm.com/Articles/Editorial/Magazine-
Features/Intelligence-in-the-Cloud-55488.aspx
Watson, H. J. (2006). BI and data warehousing in universities. Business Intelligence Journal, 11(3), 4-6.
Retrieved October 5, 2010 fromhttp://tdwi.org/research/2006/09/business-intelligence-journal-vol-11-
no-3.aspx?sc_lang=en
Biography
Dr. Stevan Mrdalj is a Professor of Computer Information Systems at
Eastern Michigan University and teaches a range of information sys-
tems courses. His research interests include business intelligence, data
mining, data warehousing and systems analysis and design. He has
over fifty articles published in journals and conference proceedings. He
has been a session chair at numerous conferences. He has been a re-
viewer for seventeen books and he has published supplements for one
textbook. He is a member of the editorial board and a reviewer for the
Computer Science and Information Systems Journal.
doc_472898674.pdf
In the last decade Business Intelligence has become one of the hottest areas in Information Technology.
Issues in Informing Science and Information Technology Volume 8, 2011
Would Cloud Computing Revolutionize Teaching
Business Intelligence Courses?
Stevan Mrdalj
Eastern Michigan University, Ypsilanti, MI, USA
[email protected]
Abstract
In the last decade Business Intelligence has become one of the hottest areas in Information Tech-
nology. Meanwhile, teaching Business Intelligence courses for Masters of Business Administra-
tion programs faces a variety of challenges. One of the major hurdles in offering Business Intelli-
gence courses is the complex and costly computing infrastructure necessary for the software used
in such courses. One possible remedy might be to use cloud computing. Cloud computing offers
flexible access the necessary software and hardware through the Internet. It is becoming increas-
ingly popular because of its on-demand and pay-per-use business models. In this paper, we exam-
ine how cloud computing might be used to revolutionize the way of teaching business intelligence
courses. In doing so, we present various possible infrastructure scenarios and the necessary tools.
Finally, we discuss the perceived difficulties in adopting Business Intelligence as a service busi-
ness model in academic settings. We also reflect on the economics of cloud computing.
Keywords: Cloud Computing, Business Intelligence, Software as a Service, Teaching BI, Pay-
per-use.
Introduction
For the last several years, Masters of Business Administration (MBA) programs have been under
mounting pressure to provide courses to prepare their students to become the next generation of
Business Intelligence (BI) workers. In a recent survey of professors from more than 80 universi-
ties around the world (Betts & Kanaracus, 2010), Barbara Wixom discovered that 43% of the re-
spondents reported that they can’t adequately teach BI courses” because they lack access to the
needed software, hardware and real-world business problems”. She also indicates that “It’s clear
from the survey that instructors want to change the way students learn about business intelli-
gence. Professors want to provide large data sets, contemporary software tools, and real-world
content within their classrooms. But, factors like high technology costs, complex maintenance
requirements, and steep learning curves present insurmountable obstacles.” With the present eco-
nomic crisis and its repercussions on the academic environments, one of the major hurdles in of-
fering Business Intelligence courses is
the complex and costly computing infra-
structure necessary for the software used
in such courses. On the other hand,
unless Business Intelligence becomes a
mainstream field at business schools, the
shortage of business intelligence related
professions will continue to rise.
Material published as part of this publication, either on-line or
in print, is copyrighted by the Informing Science Institute.
Permission to make digital or paper copy of part or all of these
works for personal or classroom use is granted without fee
provided that the copies are not made or distributed for profit
or commercial advantage AND that copies 1) bear this notice
in full and 2) give the full citation on the first page. It is per-
missible to abstract these works so long as credit is given. To
copy in all other cases or to republish or to post on a server or
The cloud computing, also known as
software-as-a-service, delivery model
has the potential to become a major
to redistribute to lists requires specific permission and payment
of a fee. Contact [email protected] to request
redistribution permission.
Would Cloud Computing Revolutionize BI Courses
force in the future of offering BI courses. After a decade of growth and adoption, the commercial
cloud computing offerings came into existence and products like Amazon EC2, Microsoft Azure
and Salesforce.com helped to popularize it. The concept of on-demand BI promises to make it
affordable for business schools to deliver quality BI instruction to MBA students.
Computer Science programs were among the first to develop experimental undergraduate data
mining courses as reported by Lu and Bettine (2003). Banks, Dong, Liu and Mandvikar (2004)
report on teaching data mining courses as “an exciting addition to the curriculum at the sen-
ior/graduate level” that provide the opportunity to apply computer related education to various
domains and applications. Paper by Roiger (2005) presents a very good tutorial on data mining
that can be used by educators as an introduction to business intelligence. The typical structure of
a data warehousing and data mining course in the Computer Science curriculum is well described
by Fang and Tuladhar (2006) and Musicant (2006). Information systems programs at business
schools typically offer BI courses as elective courses (Watson, 2006). A good example of an ap-
plied BI course that is designed to appeal to the graduate students of the CIS programs as well as
MBA students selecting the CIS concentration is presented by Mrdalj (2007). Specifically, the
structure and components of this course focus on “business aspects” of data mining and data
warehousing by having students learn how to answer today’s BI questions. In one of the rare pa-
pers related to MBA programs, Mrdalj and Diallo (2010) present the structure and components of
a BI module that is designed to appeal to students of the finance program as well as to MBA stu-
dents selecting the finance concentration.
In the next section of this paper we stage our proposed solution with a brief presentation of the
typical computing infrastructure architecture needed for teaching BI courses with a discussion of
possible on-premises installations. It is followed by the description of cloud computing and its
possibilities in teaching BI courses. We conclude with perceived difficulties in using cloud com-
puting as a platform for teaching BI courses.
Computing Requirements for BI Courses
The typical computing infrastructure architecture needed for teaching BI courses is shown in Fig-
ure 1. It requires a database server to host source databases as well as the data warehouse.
Data Warehousing
OLAP
Cubes
BI Tools
Data
Sources
INSTRUCTOR
STUDENT
ERP/
CRM
OLTP
Legacy
System
Extract
Clean
Transform
Load
Figure 1: A Typical BI Architecture
210
Mrdalj
Next, we need data warehousing development tools to extract data from the sources and to clean,
transform and load data into the data warehouse. Although this part is not visible to the MBA stu-
dents, it is required for the instructors in order to provide the students with "real-life" data sets so
they can make sure that students learn from actual company data instead of textbook examples.
In addition, BI architecture requires technology such as Online Analytical Processing (OLAP)
that provides a platform for end-user based utilization of data warehouse (Larson, 2009) and is a
dominant paradigm in BI. This technology is essential for MBA students who are usually not fa-
miliar with SQL. OLAP tools give students flexible and fast access to multi-dimensional views
(such as store, account, time, geography) and to measures (such as sales, payments, discounts),
both summarized and detailed. Such access if further supported with standard OLAP operations
that include drill-down, roll-up, dice and slice. Another benefit of using OLAP Cubes is that stu-
dents can access them with tools such as the Microsoft Excel PivotTable (Harts, 2008; Mrdalj &
Diallo, 2010) as illustrated in Figure 2.
Tools like the Microsoft SQL Server Data Mining Add-ins (MacLennan, Tang & Crivat, 2009)
and SAS
®
Enterprise Miner (Cerrito, 2006) enable Microsoft Excel to become the analytical cli-
ent to the OLAP analytical servers. This enables MBA students to access multi-dimensional data
sources and to perform reporting, ad-hoc analysis and data mining within the very familiar envi-
ronment of Microsoft Excel (Mrdalj & Diallo, 2010). Consequently, Microsoft Excel, perhaps the
oldest BI tool, has regained its role in MBA education as an affordable BI front end tool.
The above described BI architecture can be implemented on-premises in several different ways.
We will discuss the following three most common approaches.
Stand Alone Infrastructure
The simplest and most restrictive is the stand-alone approach when the entire BI architecture (da-
tabase server, analytical server and BI tools) is installed on each individual workstation in the lab
or computer classroom. It requires early setup of all the software and data sets which can be eas-
ily cloned to all workstations. To accommodate the continuity in students’ work, such setup
would require that students use the very same workstation throughout the entire semester and to
keep their own backups. Each change in software and data sets during a semester would require a
manual installation of such changes on all workstations since re-cloning would diminish the pre-
vious student’s work. This setup would also require that students would need to do their assign-
OLAP
Figure 2: End-user BI Access
211
Would Cloud Computing Revolutionize BI Courses
ments and projects on very same workstations that they use during classes or labs. Such a re-
quirement poses serious constraints on the type and quantity of assignments that can be given to
students. Otherwise, students would need to have the complete BI architecture installed on their
laptops/desktops.
Setting up a proper BI environment on students’ laptops is a multi-step process (for illustration
purposes we will describe a scenario with Microsoft SQL Server). First, it requires installing the
SQL Server. The next step is to install sample data warehouses including the corresponding
OLAP Cubes. Unfortunately, deployment of OLAP Cubes requires some familiarity with the Mi-
crosoft SQL Server Management Studio or the Microsoft Business Intelligence Development
Studio project. The next step is to download and install Microsoft SQL Server Data Mining Add-
ins. The last task of this process is to connect Excel to the SQL Server Analysis Service and to
configure an analysis database to be used by Excel to perform data mining tasks. The above de-
scribed multi-step installation process would certainly provide a major challenge to MBA stu-
dents as well as the necessity to acquire all the needed software. Based on our experience in using
this approach, we consider this to be prohibitive for the majority of MBA students and, therefore,
we consider it a most restrictive approach for teaching BI courses.
Local Area Network Infrastructure
The step-up approach would have a local area network where the database server hosting the data
warehouse and the analysis server hosting the OLAP cubes are installed on the network server
and all the client workstations would need only Excel with BI tool add-ins as shown in Figure 3.
Figure 3: Network Configuration
This approach certainly allows any mid-semester changes in the software and data sets. It requires
moderate DB administration since all clients would need proper authentication to access OLAP
cubes. The major problem with this approach is that those students wanting to do their assignment
on their laptops would need to be permanently connected to the network since they would need
access to the analysis server to perform BI tasks. Students who would like to do their work out-
side of the lab environment would need to have the stand-alone installation described above.
Web Enabled Infrastructure
To overcome the problems with the previous two approaches, business schools would need to
invest in the web enabled solution depicted in Figure 4. The advantage of this approach is that
students would only need to install Excel add-ins and to have available Internet access. On the
other hand, business schools would need considerable hardware and software resources to im-
212
Mrdalj
plement it. In addition, this approach would require a high level of expertise to configure database
and web servers as well as to make them operational non-stop. The cost of designing, building,
installing and running the web enabled infrastructure continues to be the most noticeable limiting
factor for business schools. Frequently such prohibitive costs have been the main limitation for
proliferation of BI courses at MBA programs.
Figure 4: Web Enabled Configuration
Another variation of this approach is to use remote access tools for the remote clients. Such an
approach would even eliminate the need for students to have Excel as well as BI tools installed on
their laptops. A comprehensive discussion of the remote access tools (Scher, 2010) is beyond the
scope of the paper, but we list some of them as examples: GoToMyPC, LogMeIn Pro2, Microsoft
Remote Desktop Connection, Remote PC and pcAnywhere.
If the universities do not have adequate resources or if they do not want to acquire and manage
technical resources or if they lack the necessary technical experience and skills, then they should
consider using cloud computing.
Using Cloud Computing for Teaching Business
Intelligence
Cloud computing is one of the most attractive technology areas today due to its cost efficiency
and flexibility. There are many definitions and interpretations of cloud computing, but for the
purpose of this paper we will use the following definition created by the National Institute of
Standards (Mell & Grance, 2009):
“Cloud computing is a model for enabling convenient, on-demand network access to a shared
pool of configurable computing resources (e.g., networks, servers, storage, applications, and ser-
vices) that can be rapidly provisioned and released with minimal management effort or service
provider interaction.”
In the cloud computing paradigm, the basic architecture needed for teaching BI courses based on
the three layered NIST service model (Mell & Grance, 2009) and its customizations for BI
(Chandra & Iyer, 2010; Reyes, 2010) is presented in Figure 5. At the lowest level, the processing
power refers to the hardware needed for running database servers and networking. Normally, the
customer does not control or manage this layer of the cloud infrastructure. The cloud’s Platform
as a Service (PaaS) provides instructors with the capability to deploy large data warehouses from
real businesses. The Software as a Service (SaaS) layer of the cloud provides students and in-
structors with applications running on a cloud infrastructure. These layers may be considered as
services to the layer above them.
213
Would Cloud Computing Revolutionize BI Courses
In relationship to the BI courses, the data warehousing tools would allow instructors to deploy
data warehouses and OLAP cubes as well as to maintain them over the course of the semester.
Students would gain access to the OLAP cubes and reporting, data mining, and business perform-
ance management applications. The OLAP cubes and applications are accessible from remote
clients’ devices trough a thin client interface such as a web browser (Mell & Grance, 2009) or
Excel add-ins (Predixion, 2010). In short, cloud computing shifts the physical construction of the
data center, creating network connections between students and the center, and providing all the
necessary software to the trusted service provider (Thomson, 2009).
Figure 5: Basic BI Architecture in the Cloud
Thanks to recent advances in technology in recent years, the commercial cloud computing offer-
ings may enable business schools to implement BI courses in a fraction of the time and avoid the
capital expenditure required by traditional installations (Thomson, 2009). Frequently mentioned
advantages of cloud computing are lower cost and non-stop operation. Even more important fea-
tures for the academic environments are summarized below (adapted from Katzan, 2009):
• The responsibility for hardware, application software, storage facilities, and professional
services resides with the provider.
• Systems software is available from a trusted vendor for supporting cloud services.
• Data centers are available for sustainable and reliable operation.
In other words, cloud computing would provide virtual and scalable infrastructure as a service
where the complexity of resource management is hidden from the client while enabling the client
to exploit supercomputing power on-demand without investing in huge infrastructure and man-
agement costs (Ananthanarayanan et al., 2009). A broad discussion of the available Cloud BI
tools is beyond the scope of this paper and it can be found in (Rayes, 2010; Tsai, 2009). Instead,
we list some of them as examples: Predixion Insight, CloudOLAP, IBM Smart Analytics Cloud,
Informatica 9, and Rockspace Cloud.
This brings the “thin client” architecture back to life where both instructors and students would
utilize the needed services from the cloud provided as shown in Figure 6. The biggest advantage
that cloud computing brings to instructors is that they can obtain the proof-of-concept for the de-
velopment of their courses in a matter of days. Instructors would be able to upload their example
214
Mrdalj
data warehouses and OLAP Cubes and make them available to students wherever Internet access
is available. Students would gain easy and location independent access without the need to ac-
quire complex and expensive software that needs to be installed on their workstations in order to
do their assignments.
OLAP
DW
ETL
STUDENT
INSTRUCTOR
Figure 6: Cloud BI
Perceived Difficulties in Using Cloud Computing
Despite many advantages and benefits that cloud computing brings to business schools there are
several perceived difficulties as well. In this paper, we will only concentrate on those that are
relevant to using cloud computing for teaching purposes. Many other challenges related to cloud
computing used for commercial purposes such as scalability, security, large data volume, speed of
access, and reliability can be found in (Reyes, 2010; Thomson, 2009).
Customization is one of the very important aspects of cloud computing used for teaching BI
courses. With non-configurable SaaS (Katzan, 2009), the cloud provider delivers a unique set of
applications to clients and limits instructors to tailor them to specific classwork needs. A similar
concern is for instructors teaching multiple sections when a multi-tenant SaaS would be required
to host common applications and unique data sets for different sections.
Another important consideration of adapting cloud computing for teaching BI courses is that its
pricing structure is variable and complex. There are three categories that we might observe as
valid options for the academic environments (adapted from Katzan, 2009):
• Perpetual license refers to an “up front” payment for service, and unlimited access for an
unlimited time.
• Subscription, as a form of cloud service monetization, can be conceptualized as a time-
based perpetual license, often applied to multiple users.
• Transaction based or variable pricing model, also known as pay-as-you-go plan, is based
on the data transfer volume and number of transactions.
The variable pricing model may look quite attractive for business schools since data volume and
number of transactions may not be large, but it would bring an uncertain cost associated with a
given course. Many would argue that the subscription model is more economical and that it pro-
215
Would Cloud Computing Revolutionize BI Courses
vides budgetary predictability and a lower cost over a longer period of time (Thomson, 2009).
What may drive up the cost for either pricing model are configurability and isolation. From a
pricing stand point non-configurable and single-tenant SaaS have a lower cost, but are quite pro-
hibitive in customization.
Both pricing models are something that business schools may not be accustomed to have associ-
ated with offering courses. Traditionally, universities have budgets for computer refresh programs
and laboratory maintenance and not for cloud subscription fees associated with an individual
course. Although either of the pricing models mentioned above might cost much less compared to
the costs of on-premises BI architectures, it would require a paradigm shift in university budget-
ing models.
Conclusions
We examined the requirements for implementing the BI architecture necessary for teaching BI
courses. Based on them, we concluded that cloud computing is an attractive solution for business
schools wanting to implement cost-effective, rapid and dynamic environments for their BI
courses. The promise of cloud computing is already proven at many companies and it might be
prudent for business schools to adopt this new approach instead of investing in ever more de-
manding and expensive computer labs and classrooms.
By outsourcing the physical construction of the data center, creating network connections be-
tween students and the center, and providing all the necessary software to students, business
schools can concentrate on rapid development of BI courses and their integration into MBA pro-
grams. With cloud computing, business schools may significantly reduce their costs while gaining
guaranteed performance and high student satisfaction.
We also hope that cloud-based BI service providers like Predixion, CloudOLAP and especially
Microsoft Azure (with the offering of the analysis server in the near future) would be willing to
provide academic discounts for their services to those universities teaching BI courses.
References
Ananthanarayanan, R., Gupta, K., Pandey, P., Pucha, H., Sarkar, H., Shah, M., & Tewari, R. (2009). Cloud
analytics: Do we really need to reinvent the storage stack? Proceedings of the 2009 Conference on Hot
Topics in Cloud Computing. Retrieved September 12, 2010 fromhttp://www.usenix.org/event/hotcloud09/tech/full_papers/ananthanarayanan.pdf
Banks, D. L., Dong, G., Liu, H., & Mandvikar, A. (2004). Teaching undergraduates data mining engineer-
ing programs. Proceedings of the 34th ASEE/IEEE Frontiers in Education Conference, October 20 –
23, 2004, Savannah, GA, T1A-1 - T1A-6. Retrieved September 22, 2010 fromhttp://www.public.asu.edu/~huanliu/papers/fie04.pdf
Betts, M., & Kanaracus, C., (2010, April 5). Professors cite challenges in teaching BI in college. Com-
puterworld. Retrieved November 10, 2010 fromhttp://www.computerworld.com/s/article/347880/Professors_Struggling_to_Teach_BI
Chandra, B., & Iyer, M. (2010). BI in a cloud: Defining the architecture for quick wins. SETLab Briefings,
8(1), 39-44. Retrieved October 10, 2010 fromhttp://www.infosys.com/research/publications/Documents/BI-in-a-cloud.pdf
Cerrito, P. B. (2006). Introduction to data mining using SAS Enterprise Miner. Cary, NC: SAS Institute.
Fang, R., & Tuladhar, S. (2006). Teaching data warehousing and data mining in a graduate program of in-
formation technology. Journal of Computing Sciences in Colleges, 21(5), 137-144.
Harts, D. (2008). Microsoft Office 2007 business intelligence: Reporting, analysis, and measurement from
the desktop. McGraw-Hill/Osborne.
216
Mrdalj
217
Katzan, H. (2009). Cloud software service: Concepts, technology, economics. Service Science, 1(4), 256-
269.
Larson, B. (2009). Delivering business intelligence with Microsoft SQL Server 2008. McGraw-
Hill/Osborne.
Lu, Y., & Bettine, J, (2003). Data mining: An experimental undergraduate course. Journal of Computing
Sciences in Colleges, 18(3), 81-86.
MacLennan, J., Tang, Z. H. & Crivat, B. (2009). Data mining with SQL Server 2008. Wiley Publishing.
Mell, P., & Grance, T. (2009). The NIST definition of cloud computing. National Institute of Standards and
Technology, Version 15, 10-7-09. Retrieved October 10, 2010 fromhttp://csrc.nist.gov/groups/SNS/cloud-computing/
Mrdalj, S (2007). Teaching an applied business intelligence course. Issues in Information Systems, 8(1),
134-138.
Mrdalj, S., & Diallo, A., (2010). Bringing business intelligence into finance curriculum. Issues in Informa-
tion Systems, 11(2), 168-175.
Musicant, D. R. (2006). A data mining course for computer science. ACM SIGCSE Bulletin, 38(1), 538-
542.
Predixion Insight. (2010). Predixion software. Retrieved November 15, 2010 fromhttps://www.predixionsoftware.com/predixion/Portals/0/pdf/Predixion_Insight_DS_FINAL_9_4_10.p
df
Reyes, E.P. (2010). A systems thinking approach to business intelligence solutions based on cloud comput-
ing. Boston, MA: Massachusetts Institute of Technology. Retrieved November 2, 2010 fromhttp://hdl.handle.net/1721.1/59267
Roiger, R. J. (2005). Teaching an introductory course in data mining. ACM SIGCSE Bulletin, 37(3), 415-
421.
Scher, R., (2010, October). Remote access tools. PC Today, 20-23.
Thompson, J. (2009). Business intelligence in a SaaS environment. Business Intelligence Journal, 14(4),
50-55. Retrieved October 5, 2010 fromhttp://tdwi.org/research/2009/12/business-intelligence-journal-
vol-14-no-4.aspx?tc=page0
Tsai, J. (2009, August). Intelligence in the cloud. Customer Relationship Management Magazine, 38-43,
Retrieved October 15, 2010 fromhttp://www.destinationcrm.com/Articles/Editorial/Magazine-
Features/Intelligence-in-the-Cloud-55488.aspx
Watson, H. J. (2006). BI and data warehousing in universities. Business Intelligence Journal, 11(3), 4-6.
Retrieved October 5, 2010 fromhttp://tdwi.org/research/2006/09/business-intelligence-journal-vol-11-
no-3.aspx?sc_lang=en
Biography
Dr. Stevan Mrdalj is a Professor of Computer Information Systems at
Eastern Michigan University and teaches a range of information sys-
tems courses. His research interests include business intelligence, data
mining, data warehousing and systems analysis and design. He has
over fifty articles published in journals and conference proceedings. He
has been a session chair at numerous conferences. He has been a re-
viewer for seventeen books and he has published supplements for one
textbook. He is a member of the editorial board and a reviewer for the
Computer Science and Information Systems Journal.
doc_472898674.pdf