Key Requirements for Predictive Analytical IT Service Management
Architectural Key Characteristics for a Cloud based Realization
Christopher Schwarz, Han
s Peter Bauer, Lukas Blödorn and Erwin Zinser
Institute of Information Management, FH Joanneu, Gesellschaft mbH, Alte Poststrasse 147, Graz, Austria
Keywords: IT Service Management, Predictive Analytics, Business Analytics, Service Oriented Architecture, Service
Bus, Semantic Web Technologies, Semantic Reasoning, Controlled Natural Language, Cloud Computing.
Abstract: While trying to maintain sustainable competitive advantage, IT service providers are challenged with
tremendous service complexity and a low level of flexibility caused by the lack of transparency, constrained
scalability and the missing ability to identify needed service measures proactively. For overcoming these
challenges, this paper presents a well-evaluated set of identified key requirements for a feasible realization of
a highly scalable cloud based architecture that supports predictive analytics in several domains of IT Service
Management. This presented concept goes far beyond traditional approaches and pertinent state-of-the-art
software solutions by focusing on business analyses based on knowledge creation and domain-independent
knowledge sharing. The proposed approach is based on profound analyses of related work as well as modern
service oriented design and business analyses paradigms. It provides semantic complexity handling, structured
and multi-layered service interaction, cloud-enabled scalability management as well as predictive business
analyses based on semantic reasoning, decision-making support and pattern recognition. The derived results
eventually provide solution architects with a feasible and technical independent fundament for architectural
implementation decisions. It ultimately enables IT service providers to cope with modern flexibility needs
and complexity challenges and therefore to continuously satisfy customers to gain competitive advantage.
1 INTRODUCTION
A company's success fundamentally depends on its
ability to develop sophisticated solutions, perfectly
fitting to customers’ requirements. Nevertheless, to
guarantee sustainable success, bearing up customers'
satisfaction is necessary, especially if requirements
change. Hence, a flexible and agile service network is
needed, adaptable to changes, but most of all
adaptable to customers’ requirements.
According to Porter’s Five Forces, competitive
rivalry in the IT area is significantly higher than in
many other branches, primarily caused by a high
threat of new entry and substitution (Fung, 2013, pp.
19,20). The internet and cloud computing eliminate
many physical barriers of entering a new market
(Fung, 2013, pp. 19,20). Due to the variety of IT
services with similar functionality it is easy for
customers to change to substitutes, whereas IT
service providers have to deal with a high range of
competitive companies, serving those customers
(Fung, 2013, pp. 19,20). Thus, albeit it is
comparatively easy to provide IT services, it requires
a lot of effort to keep up with others by continuously
satisfying customers and managing their
requirements not just reactively, but in a proactive
and predictive way for effective decision-making.
The ability for successfully managing services
over a long-time period distinguishes an effective
from an ineffective service provider. The key to
success is flexibility in IT Service Management
(ITSM), by constantly aligning and adjusting the
service offers to the actual market need. Managing
services means managing a network of
interdependent components that have to be
completely coordinated. Disability of coping with
multitenant service dependencies leads to missing
transparency in the service environment, which
affects calculation and billing of services and makes
analytics for business relevant financial decisions
extremely difficult, or even impossible (Schwarz, et
al., 2013, p. 1).
The strong influence of ITSM on Financial
Management indicates the need for balance and
flexibility over multiple domains, which is still an
unhandled problem of current software solutions
297
Schwarz C., Bauer H., Blödorn L. and Zinser E..
Key Requirements for Predictive Analytical IT Service Management - Architectural Key Characteristics for a Cloud based Realization.
DOI: 10.5220/0005490502970303
In Proceedings of the 5th International Conference on Cloud Computing and Services Science (CLOSER-2015), pages 297-303
ISBN: 978-989-758-104-5
Copyright
c
2015 SCITEPRESS (Science and Technology Publications, Lda.)
(Schwarz, et al., 2013, p. 1). IT is the most pervasive
factor in business, affecting every single process and
decision (Addy, 2007, p. 20). Therefore the discipline
of ITSM is applicable for nearly all business domains
relying on any kind of IT based service. Moreover, it
is necessary to control all IT components from a
holistic view, for constantly aligning the business
strategy with the IT strategy, especially when
environmental changes appear (vom Brocke, et al.,
2013, p. 1).
Nevertheless, complexity is considered as the
major challenge in ITSM. Complexity is referred to
the management of service data and its relations
(Benedettini and Neely, 2012, pp. 5,6).
Interdependencies and data flows between services
have to be determined easily and fast, so that possible
effects of changes and events can be recognized
ideally in real-time. Thus, efficient ITSM requires
transparency of service-to-customer, as well as
service-to-service interaction. Besides that,
transparency is necessary between services and its
underlying components, to ensure appropriate service
functionality. In the same way profound knowledge
of service relations is needed for defining and
fulfilling cost-efficient service levels and
consequently to ensure a particular Quality of Service
(QoS) level. Constantly changing requirements
necessitate QoS management techniques that are far
beyond static provisioning of network resources
(Kourtesis, et al., 2014, pp. 306,307). Higher
elasticity and dynamic provisioning is needed, based
on real-time decisions, reasoning and inference
(Kourtesis, et al., 2014, pp. 306,307).
Second, the maintenance and continuous
improvement of a service infrastructure, tightened
influenced by IT outsourcing and cloud-oriented
design, define a new level of complexity (Benedettini
and Neely, 2012, pp. 5,6). Changing requirements
often demand changing functionality, seamlessly
integrated in the working service environment.
Up to a certain level, IT services seem to be
manageable easily. With growing complexity,
however, a point will be reached where the effort of
complexity handling is higher than the services’
benefit, leading to crucial cost inefficiency (Josuttis,
2007, p. 2). Consequently, the service infrastructure
has to be designed and structured in a way that
supports effective management of both mentioned
complexity types on the one hand and enables
seamless integration of new services on the other
hand. The high level of complexity forces IT service
providers to think of new ways for automatically
managing a variety of data, variables and parameters,
necessary for the operation of services and its
resources (Kourtesis, et al., 2014, p. 308).
2 RELATED WORK
In the past years, different approaches have been
presented in the area of ITSM, all with the overall
goal to enhance efficiency, primary by using the
Semantic Web concept. They provide sophisticated
solutions with a precise goal for dealing with
complexity in IT Service Management.
It is beyond doubt that these solutions provide
value in the very specific field they are used, but the
question arising is, if they are adaptable to other
correlating areas as well at an adequate level of effort.
Referring back to the de-facto ITSM standard, the
Information Technology Infrastructure Library
(ITIL), IT Service Management is defined as the
discipline to deal with all processes in the service
lifecycle (van Bon, et al., 2007, pp. 24-26, 42).
Efficient ITSM needs a holistic view, accomplishing
a platform that allows making use of the Service
Oriented Architecture (SOA) and analytical benefits
in any ITSM domain. The ability of using semantic
knowledge must not be limited to one specific
implementation, but has to be realized on a shared
base. The semantic Wiki-based approach of Kleiner
et al. (Kleiner, et al., 2012) perfectly fits to the issue
of complexity in Incident and Problem Management
and thus, it is closely aligned to exactly these ITIL
processes. Although the semantic Wiki allows the
integration of other applications, the platform is still
limited to the information stored in the semantic
Wiki. Jantscher et al. (Jantscher, et al., 2014) focus on
reducing negative business impacts caused by wrong
incident prioritization. They developed an example
for analytical ITSM, the Incident Prioritizer, which is
an AHP decision support system for the prioritization
of incidents based on their business impact. For the
prioritization process, relevant incident data is
provided by an ontology, defining an ITIL-compliant
service catalogue. Valiente et al. (Valiente, et al.,
2012) deal with the service management problem of
integrating service management processes, which are
often specified in natural language. The paper aims to
translate ITSM relevant information, expressed in
natural language, to a computer understandable
format, by using semantic technologies. Thus, Onto-
ITIL is presented in this work, as an OWL based
ontology, tailored to be used in ITSM, to overcome
the gap between natural language process definitions
and IT services, or more generally the gap between
business and IT. Otherwise, the very generic
approach of Freitas et al. (Freitas, et al., 2008) defines
CLOSER2015-5thInternationalConferenceonCloudComputingandServicesScience
298
insufficient structure do be applicable in this stage for
operative use. The idea of a generic ontology, usable
for different domains, has definitely potential in
theory, but not enough alignment to the ITIL
processes for effective ITSM.
What is needed is an overall platform allowing the
seamless integration of ITIL process solutions
throughout the whole service lifecycle. The platform
architecture has to be designed to support flexibility
and to make services adaptable for changes. The
approach of El-Gayar et al. (El-Gayar & Deokar,
2013) with its distributed model environment
presents the key feature of using a service bus for
enabling flexible changing of distributed models,
based on specific problems.
Extending this approach of high flexibility and
easy sharing and reusing of information to the field of
ITSM, a platform can be realized that allows the
integration of process solutions throughout the whole
lifecycle and sharing and using knowledge in the
whole environment.
3 RESEARCH QUESTIONS AND
OBJECTIVES
When focusing on the implementation of existing
concepts in the related work, the need for further
development has been identified, missing a holistic
approach for seamlessly integrating them. Thus, the
presented approach in the upcoming sections tries to
answer the following research questions:
What are the key requirements for a scalable
architecture to support predictive analysis in
ITSM to be able to cope with the complexity of
service interdependencies and heterogeneity as
well as the lack of transparency?
Is there a way for seamlessly integrating services
and making use of provided functionality and
commonly used data?
Is it possible to provide a convenient way for
scalability and extensibility management that
allows flexible and on-demand use of resources?
4 PROPOSED APPROACH
In the first step, a long-term evaluation process was
necessary to identify the major characteristics for
predictive-analytical ITSM. They will be introduced
in this paper as the eight key requirements of IT
Service Management as they combine structure and
process-oriented service management of ITIL, the
architectural advantages of a SOA and the centralized
integration of semantic technologies for handling
service complexities. They have been identified as
follows:
1. Structured Service Interaction
2. Centralized Service Orchestration
3. Multi-layered Software Architecture
4. Scalable Computing Architecture
5. Domain-independent Architecture
6. Common Information Integration
7. Predictive Analyses Integration
8. Natural Language Interface
These key requirements do not specify any
technological implementations and thus provide a
technology-independent, holistic view on an ITSM
concept, tailored to be generic but structured, scalable
and extensible, and applicable for several domains of
ITSM. This overall concept provides an ITSM
environment highly adjustable to any business needs
for a clear alignment to the business processes and a
strong focus on shared processible knowledge
throughout the whole environment. The motivation
for defining these eight key requirements is to provide
a common foundation for the development of any
ITSM approach with a focus on effective and
predictive analytics. This foundation is not just
applicable for dealing with one particular ITSM
problem, but constitutes helpful practices at the
starting point of any effective ITSM development.
4.1 Structured Service Interaction
Extensive service interaction is an indicator for a
well-structured and working service environment.
Each service provides defined functionality that can
or has to be used by other services. Thus, service
communication is inevitable for requesting and
returning service provided information. Service
interaction must not be avoided. It is an indicator for
sophisticated capsulation of functionality and
conforms to the concept of information sharing and
reusing. The only condition for effective service
interaction is to accomplish a structured and
consistent way of communication. Communication
between humans works as long as they understand
each other. The same obtains for communication
between services.
However, a network of services often consists of
heterogenic technologies for service development. It
is nearly impossible, or just manageable with high
effort, to keep a service environment homogeneous,
which is not the goal for effective service
KeyRequirementsforPredictiveAnalyticalITServiceManagement-ArchitecturalKeyCharacteristicsforaCloudbased
Realization
299
communication. Rather, it is necessary, in a network
of various services, to define and enforce a
communication standard valid for all service
interaction.
In combination with a generic language, the
service interaction has to be structured with defined
communication endpoints. Interfacing the
communication ensures that information exchange is
consistent in the whole architecture, which facilitates
the maintenance and extension of the service
environment and supports the Service Transition
phase. Using consistent communication interfaces, it
is clearly defined how new services have to provide
and how to retrieve information from others, which
reduces the risk of incompatibilities in the Release
Management.
4.2 Centralized Service Orchestration
Well-defined interfaces for service requests and
responses are inevitable for service consistency and
service integration and maintenance, but they do not
prevent service networks of reaching a level of
unmanageable complexity. Service communication
has to be orchestrated over a centralized
communication manager, a service bus, to provide a
single point of contact for all service communication.
Instead of directly addressing, services contact the
centralized service bus, which handles the further
processing of the service messages, based on a
consistent and clear addressing schema for services.
However, effective communication management
is not just relaying messages from service A to service
B, without considering possible service downtimes or
overutilization. Effective communication
management has to accept the responsibility of
managing message queuing and load balancing, to
ensure a stable environment that can dynamically
react on service failure. Hence, the service bus can
perform message forwarding, without knowledge of
the actual service location. The message sender and
receiver can be located anywhere, as long as they
reach the service bus communication interfaces,
which enables the possibility for changing service
location, for instance a transformation to the cloud,
without losing connectivity to the service
environment. This flexibility in service providing, in
combination with low maintenance effort, makes
complex service networks manageable, even if
process requirements change. Decisions for
outsourcing of service functionality do not depend on
the service interdependencies anymore and can be
performed completely based on cost and compliance
reasons.
4.3 Multi-layered Software
Architecture
Functionality has to be separated into services to
provide a manageable structure and to be flexible for
changes. Basically, service functionality comprises
the ability to store and retrieve data, to process the
data, if needed based on workflows, and to present the
processed information, which can be described as the
four service functionality layers. For sure, these
layers can be developed for each service
independently, but a structured separation in a
standardized layer design, based on interface
connectivity definitely supports the architecture’s
structure and consistency and prevents unnecessary
heterogeneity. If all services store and retrieve data
based on the same standardized platform for data
storage, maintenance of service data is also limited to
this platform and does not require skills in multiple
technologies.
In addition, the integration of new services or the
replacement of service functionality can be
performed with lower risk of incompatibilities, if
storage communication is accomplished over
specified interfaces to a standardized storage
platform. On the same way, service logic and service
processes have to be implemented. Referring back to
the structured and consistent service interaction,
communication to the data storage and to other
services over the service bus has to be accomplished
over standardizes interfaces, independent from the
logic technology.
The coordination of providing information
through the data, logic and process layers, and the
presentation of this information, in the presentation
layer, is essential for bringing the service value to the
customer. The challenge of this layer is to retrieve
information over the services bus and to bring it into
a specific view. The clear separation between
information processing and information presentation
allows the interaction of different presentation views
with one underlying service module and vice versa.
This structure comes up to the service catalogue
concept, which allows the combination of service
items to different service packages, adjusted to a
customer’s need and flexible for change. Besides that,
this clear separation and the communication
management of the service bus, allow the easy
development of a presentation view for different
devices, without the necessity of changing the service
logic. The described aspects enable value-focused
development of presentation views tailored to the
customers’ needs in all aspects, collecting the
information needed and presenting it in the most
CLOSER2015-5thInternationalConferenceonCloudComputingandServicesScience
300
convenient form. The outcome can be a mobile
application for a maintenance worker or a classic
desktop program for the controller, both accessing
collected and tailored information.
4.4 Scalable Computing Architecture
Maintaining and improving an ITSM environment
demands flexibility in scalability management.
Dynamic reaction on changes is required, at best close
to real-time, to ensure service operation without any
difficulty. Thus, cloud platforms can be used, which
allow dynamic on-demand resource allocation and
flexible scalability as well as location independent
service communication provided by a service bus.
Moreover, the layered design of a service allows
the cloud-transformation of each layer independently
from the other layers. Consequently, the service data
storage could be transferred to the cloud, while the
service logic is still located on premise. This layer
based service design enables scalability management
at component level and thus, provides the maximum
of flexibility in resource provisioning. The emerging
“Big Data” topic necessitates the ability for handling
large datasets and vast amount of data for supporting
predictive decision processes by extracting the
maximum value of data (Kourtesis, et al., 2014, p.
310). Thus, highly scalable systems are needed to
process such large volumes of data in real-time
(Kourtesis, et al., 2014, p. 310).
Besides the advanced scalability management, the
support for various presentation devices is a key
feature of this cloud-enabled design. The driver for
cloud integration is not just improved scalability
management and multi device support. Moreover,
elementary, financial and compliance aspects play a
major role for switching to cloud resources. Thus, it
is important that the ITSM architecture has the
potential to switch to cloud resources with low effort
and low risk of failure. In an effective ITSM
environment, cloud decisions should only depend on
financial and compliance aspects, but definitely must
not be dependent on the technical ability to switch to
the cloud.
4.5 Domain-independent Architecture
Ontology models are, like all models, limited to a
specific area, domain or region. But for providing an
efficient and holistic ITSM environment, limitations
of ontology models and consequently limitations of
knowledge are not eligible. Since it is not possible to
define a model without boundaries, the ITSM
environment has to provide the possibility to define
various ontologies or service modules, respectively,
representing all domains of expertise in ITSM.
Nevertheless, a centralized administration of multiple
models is nearly impossible, because the definition of
each ontology model requires profound knowledge in
this specific area as well as high maintenance effort.
Therefore, a decentralized approach is needed,
allowing clients to define ontology models on their
own and sharing the reusing semantic knowledge
over the service bus. Consequently, a system is
required, which overtakes the management of the
ontology creation, the update of ontologies and rules
and the querying of ontologies.
4.6 Common Information Integration
Service functionality depends on processing data. As
already mentioned, service related data has to be
stored in the service data layer. Each service has its
own specific data, only accessible by the service
itself, independently from other services’ data.
Basically, all information needed by a service can be
stored independently and separated from others, but
this strict separation has one disadvantage. In an
ITSM environment, many services depend on
information that is related to the field of ITSM in
general and commonly used. Thus, a strict separation
of all service data leads to multiple storage of the
same information, which is unnecessary. A better
approach is to divide service related information and
common information, accessible by all services.
Besides the prevention of multiple data storage, a
centralized common data library provides the
possibility of effective maintenance and
improvement of commonly used information without
changing each service implementation.
4.7 Predictive Analysis Integration
For effective IT Service Management, an architecture
that allows flexible collecting and tailoring of
information for a specific customer is definitely
needed and plays a major role for the successful value
creation of a service. Nevertheless, the ability to
provide information does not imply that the given
information is useful for a specific purpose. It is not
the ability of providing information, but the ability of
providing knowledge that makes ITSM powerful -
knowledge in the sense of unknown and implicit
information and its combination and classification
based on rules. Furthermore, this kind of knowledge
defines a new level of predictiveness by solving
complex dependency constructs and revealing
behavioral patterns and further provides the base for
KeyRequirementsforPredictiveAnalyticalITServiceManagement-ArchitecturalKeyCharacteristicsforaCloudbased
Realization
301
proactive pattern recognition processes.
This new level of knowledge creation supports
decision-making processes based on advanced
analyses and the integration in decision support
systems. Therefore both, service logic and service
presentation need access to computable knowledge
they can process, in other words, access to semantic
information, which provides predictive knowledge,
based on inference and reasoning. By using semantic
technologies, the information value provided by
services is extended to a maximum. A maintenance
worker can definitely perform more efficient, if the
given information reveals unknown relations and
supports problem solving, not just reactively but also
proactively, based on semantically processed
analysis, pattern recognition and decision support
systems.
Providing predictive knowledge and analyses in
form of semantic ontology processing is a key
requirement for effective ITSM and has to be
permanently available for all services modules and
presentations. Thus, a connection to the service bus
has to be arranged, which allows all services in the
service environment to access semantic information.
The semantic ontology is also defined as a service,
providing the ability for other services to query
knowledge of a specific domain.
4.8 Natural Language Interface
Predictive knowledge is the key for all advanced
business and service analyses and consequently for
relevant decision making processes in any ITSM
domain. Thus, central availability of knowledge was
already defined as one of ITSM core characteristics.
Nevertheless, providing the formal representation of
semantic knowledge is not applicable in a multi
domain ITSM environment. Knowledge has to be
detached from the technology behind, by providing
Controlled Natural Language (CNL) interfaces,
allowing non-technical users to retrieve pure
knowledge independent from formal language
expressions. Predictive knowledge on-demand is the
process of transforming natural language query
statements of domain specialists into a query
language for RDF like the SPARQL Protocol And
RDF Query Language (SPARQL), returning
knowledge, processible for analyses and decision
making processes.
5 CONCLUSIONS
Flexibility is the major goal and complexity the major
challenge of IT Service Management to continuously
satisfy customers and to gain competitive advantage.
ITIL, SOA, CNL and the capability for predictive
analyses play a major role for effective ITSM, but
have to be applied correctly to disclose their full
potential, which is still unhandled sufficiently in the
related work. Hence, this paper identifies the key
requirements for effective and predictive analysis in
ITSM, to increase the level of flexibility and make
complexity manageable and knowledge available on-
demand. This set of requirements is defined from a
technical independent view, as a profound and
generally feasible fundament for architectural
implementation decisions regarding a holistic and
scalable predictive-analytical ITSM approach. The
consequent step is the conceptual realization of a
sophisticated and detailed architectural design, based
on these key requirements, including technical details
and the implementation process.
REFERENCES
Addy, R., 2007. Effective IT Service Management - To ITIL
and Beyond!. Berlin Heidelberg: Springer.
Benedettini, O. & Neely, A., 2012. Complexity in services:
an interpretative framework. Chicago, s.n.
Breitman, K., Casanova, M. A. & Truszkowski, W., 2007.
Semantic Web: Concepts, Technologies and
Applications. London: Springer.
El-Gayar, O. & Deokar, A., 2013. A semantic service-
oriented architecture for distributed model management
systems. Decision Support Systems (55), 4.pp. 374-384.
Freitas, J., Correia, A. & Brito e Abreu, F., 2008. An
Ontology for IT Services. Gijón, s.n.
Fung, H. P., 2013. Using Porter Five Forces and
Technology Acceptance Model to Predict Cloud
Computing Adoption among IT Outsourcing Service
Providers. Internet Technologies and Applications
Research, pp. 18-24.
Jantscher, M., Schwarz, C. & Zinser, E., 2014.
Development of an innovative AHP-based decision
support system in the field of IT Service Management.
Washington D.C., s.n.
Josuttis, N. M., 2007. SOA in Practice: The Art of
Distributed System Design. Sebastopol, CA: O'Reilly
Media, Inc..
Kleiner, F., Abecker, A. & Mauritczat, M., 2012. Incident
and Problem Management using Semantic Wiki-
enabled ITSM Platform. Vilamoura, s.n.
Kourtesis, D., Alvarez-Rodríguez, J. M. & Paraskakis, I.,
2014. Semantic-based QoS management in cloud
systems: Current and future challenges. Future
Generation Computer Systems (32), pp. 307-323.
Krafzig, D., Banke, K. & Slama, D., 2005. Enterprise SOA:
Service-oriented Architecture Best Practices. New
Jersey: Prentice Hall Professional.
CLOSER2015-5thInternationalConferenceonCloudComputingandServicesScience
302
Schwarz, C., Schmidt, S., Sellner, A. & Zinser, E., 2013.
Service Oriented Cost Accounting - Utilization-based
accounting and charging of IT service costs. Los
Angeles, s.n.
Schwitter, R., 2010. Controlled Natural Language for
Knowledge Representation. Beijing, s.n., pp. 1113-
1121.
Valiente, M.-C., Garcia-Bariocanal, E. & Sicilia, M.-A.,
2012. Applying an ontology approach to IT service
management for business-IT integration. Knowledge-
Based-Systems (28), pp. 76-87.
van Bon, J. et al., 2007. Foundations in IT-Service-
Management Based on ITIL v3. Zaltbommel: Van
Haren Publishing.
vom Brocke, J., Braccini, A. M., Sonnenberg, C. &
Spagnoletti, P., 2013. Living IT infrastructures - An
ontology-based approach to aligning IT infrastructure
capacity and business needs. International Journal of
Accounting Information Systems.
Wang, Y. H., Cao, K. & Zhang, X. M., 2013. Complex
event processing over distributed probabilistic event
streams. Computers and Mathematics with
Applications (66), pp. 1808-1821.
KeyRequirementsforPredictiveAnalyticalITServiceManagement-ArchitecturalKeyCharacteristicsforaCloudbased
Realization
303
