Towards a Framework for Enterprise Architecture
Analytics
Rainer Schmidt,
Michael Möhring
Munich University
Munich, Germany
Rainer.Schmidt@hm.edu,
Michael.Moehring@htw-aalen.de
Matthias Wißotzki,
Kurt Sandkuhl
University of Rostock
Rostock, Germany
Dirk Jugel,
Alfred Zimmermann
Reutlingen University
Reutlingen, Germany
Abstract—
Current approaches for enterprise architecture lack analytical
instruments for cyclic evaluations of business and system
architectures in real business enterprise system environments.
This impedes the broad use of enterprise architecture
methodologies. Furthermore, the permanent evolution of systems
desynchronizes quickly model representation and reality.
Therefore we are introducing an approach for complementing the
existing top-down approach for the creation of enterprise
architecture with a bottom approach. Enterprise Architecture
Analytics uses the architectural information contained in many
infrastructures to provide architectural information. By applying
Big Data technologies it is possible to exploit this information and
to create architectural information. That means, Enterprise
Architectures may be discovered, analyzed and optimized using
analytics. The increased availability of architectural data also
improves the possibilities to verify the compliance of Enterprise
Architectures. Architectural decisions are linked to clustered
architecture artifacts and categories according to a holistic EAM
Reference Architecture with specific architecture metamodels. A
special suited EAM Maturity Framework provides the base for
systematic and analytics supported assessments of architecture
capabilities.
Keywords—Enterprise Analytics, Enterprise Architecture
I. INTRODUCTION
Enterprise Architecture represents the overall structure of the
enterprise, which is composed of its business and IT structures,
like stakeholders, strategy, business capabilities, domains and
functions, business and IT processes, business products,
business services, IT services, IT applications, and technologies.
The quality and completeness of information, however,
decreases when going from top to bottom. The top layers of
architecture models contain complete and up-to-date
information. This changes for lower-level information such as
concrete IT services and applications. This information is
hitherto difficult to collect and especially difficult to keep up-to-
date.
Therefore it does not surprise that current approaches for
enterprise architecture lack analytical instruments for cyclic
evaluations of business and system architectures in real business
enterprise system environments. The aim of our research is to
close this gap and enhance analytical instruments for cyclic
evaluations of information systems and enterprise architectures
in real business environments.
On the other hand, modern infrastructure systems contain a
lot of information that describes architectures on a low
abstraction-level. This is due to the broad use of virtualization
[1]. All items of a virtualized environment are completely
described in associated management systems like configuration
management systems[2]. However, the use of these data for
enterprise architecture analytics had been hampered by
shortcomings of information technology, limiting the volume,
variety and velocity of data collection and analysis.
This has changed since the advancement of Big Data. Big
Data can be best understood as a transition integrating several
technological developments [3]. First, the management of data
has become much more powerful. Today, it is possible to
distribute even relational databases world-wide [4]. Second, new
approaches for data processing based on commodity hardware
allow to handle the processing of very large volumes of data
such as Hadoop [5]. Third, data are processed increasingly as
stream enabling decision making in real-time or near real-time.
Using technologies from the Big Data context it is possible
to complement the architectural information by a bottom-up
perspective. Data collected from infrastructure systems,
processed and analyzed using Big Data technologies is possible
to provide valuable additional architectural information.
Enterprise Architectures may be discovered, analyzed and
optimized. Furthermore the compliance of Enterprise
Architectures may be verified.
In this paper we introduce a framework for enterprise
architecture analytics, which we are integrating from an
extended service-oriented enterprise architecture reference
model in the context of Big Data analytics for architecture, new
decision support methods for architecture alignment, and an
original architecture maturity approach.
The paper is organized into the following sections. First,
suitable abstracted Enterprise Architecture models and views are
clustered using the Enterprise Software Architecture Reference
Cube to provide a normative classification base for EA
Analytics considering architectural artifacts, which are prepared
for associated decisions and architecture lifecycle information.
Then the key concepts Enterprise Architecture Analytics are
defined. To do the new potential for Enterprise Architecture
created by Big Data and Business Analytics are discussed. An
original EA Maturity Framework provides an example for using
analyses created by our Enterprise Architecture Analytics
approach. After discussing related work we conclude with
intermediate results from our research showing also the ideas for
future work.
II. ENTERPRISE SERVICES REFERENCE ARCHITECTURE
In areas where flexibility or agility in business is important,
services computing is the approach of choice to organize and
utilize distributed capabilities for Cloud and Big Data
applications. Innovation oriented companies have introduced in
recent years service-oriented architectures to assist in closing the
business - IT gap and making it cloud-ready. The benefits of
SOA are recognized for systems on the way to cloud computing
and being ready for extended service models. They comprise
flexibility, process orientation, time-to-market, and innovation.
The OASIS Reference Model for Service Oriented
Architecture [6] is the basic service-oriented abstract
architecture framework, which guides the construction of
Reference Architectures, like [7] [8]. We are using the
fundamental concepts and definitions for Software Architecture,
Architecture Reference Model, and Reference Architecture from
[9], and expand these for our research on Service-oriented
Enterprise Architectures, as in [10], [11], [12].
Fig. 1. ESARC - Enterprise Software Architecture Reference Cube.
The ESARC – Enterprise Services Architecture Reference
Cube [13] (Fig. 1) is an enterprise reference architecture model,
which completes architectural standards from [14] and [15] for
Service-oriented EAM – Enterprise Architecture Management.
We have integrated additionally fundamental concepts and
architectures for Services Computing, like in [16] [17] [18], as
well as architecture references of Cloud Computing from [19]
and [20] , as well as current Cloud Reference Architectures [21]
[22] [23] [24]. ESARC is our still growing original Service-
oriented Enterprise Architecture Reference Model, which
provides an integral EAM model for main interweaved
architectural viewpoints. ESARC abstracts from a concrete
business scenario or from specific technologies.
The Open Group Architecture Framework [14] provides
together with the current standard of ArchiMate [15] the basic
blueprint and structure for our extended service-oriented
enterprise architecture domains like: Architecture Governance,
Architecture Management, Business and Information
Architecture, Information Systems Architecture, Technology
Architecture, Operation Architecture, and Cloud Services
Architecture. ESARC provides a coherent aid for clustering,
classification, examination, comparison, quality evaluation and
optimization of enterprise architectures.
To be able to integrate architectural resources from the state
of art and practice we have developed the Enterprise Services
Architecture Metamodel Integration – ESAMI [25], as a
correlation-based integration method for architecture
viewpoints, views and models. The following few examples of
interrelated reference architectures of ESARC are the result
from correlation-based mappings of architectural models and
their elements. The Business and Information Reference
Architecture – BIRA provides a single source and
comprehensive repository of business-related knowledge from
which concrete corporate initiatives will evolve and link. This
knowledge is model-based and defines an integrated enterprise
business model, which includes organization models and
business processes. The BIRA opens a connection to IT
infrastructures, IT systems, and software as well as security
architectures.
We are using metamodels to define architecture model
elements and their relationships within ESARC [13] and [25].
We use metamodels as an abstraction for architectural elements
and relate them to architecture ontologies [26]. Architecture
ontologies represent a common vocabulary for enterprise
architects who need to share their information based on
explicitly defined concepts. Ontologies include the ability to
automatically infer transitive knowledge. The Metamodel of the
Business & Information Reference Architecture – BIRA
consists of ESARC-specific concepts, which are derived as
specializations from generic concepts such as Element and
Composition from the Open Group’s SOA Ontology [26].
The ESARC Information Systems Reference Architecture –
ISRA is the application reference architecture and contains the
main application-specific service types, defining their
relationship by a layer model of building services. The core
functionality of domain services is linked with the application
interaction capabilities and with the business processes of the
customer organization. In our research we are integrating
reference architecture models for services computing [6] [7] [8],
and extend them for cloud computing.
In the ESARC – Information Systems Reference
Architecture we have differentiated layered service types. The
information services for enterprise data can be thought of as data
centric components, providing access to the persistent entities of
the business process. The capabilities of information services
combine both elementary access to CRUD (create, read, update,
delete) operations and complex functionality for
finding/searching of data or complex data structures, like data
composites or other complex-typed information. Close to the
access of enterprise data are context management capabilities,
provided by the technology architecture: error compensation or
exception handling, seeking for alternative information,
transaction processing of both atomic and long running and
prevalent distributed transactions.
Cloud architectures are still under development and have not
reached today their full potential of integrating EAM with
Services Computing and Cloud Computing. The ESARC –
Cloud Services Architecture provides a reference-model-based
synthesis of current standards and reference architectures, like
[21] [22] [23] [24]. The NIST Cloud Computing Reference
Architecture [21] defines the Conceptual Reference Model
from the perspectives of the following Actors in Cloud
Computing: Cloud Consumer, Cloud Provider, Cloud Auditor,
and the Cloud Broker. The NIST standard defines following
deployment models: private cloud, community cloud, public
cloud, and hybrid cloud. Cloud Computing offers essential
characteristics like: on-demand self-services, broad network
access, resource pooling, rapid elasticity, and measured
services. The fundamental part of the NIST Reference
Architecture is defined by following Cloud Service Models:
IaaS – Infrastructure as a Service, PaaS – Platform as a Service,
and SaaS – Software as a Service. Some Standard extensions
like [22] provide practical additions for supporting more
directly modern business architectures by BPaaS – Business
Process as a Service and by giving a direct link by integrating
with business services to a service-oriented Enterprise
Architectures. Security additions from the CSA Security
Guidelines for Critical Areas of Focus in Cloud Computing [23]
defines a Jericho-Security-focused Service-oriented Reference
Architecture for Cloud Computing and integrates the
management perspectives from standards like ITIL and
TOGAF. The Service-Oriented Cloud Computing (SOCCI)
Framework [24] is an enabling framework for an integrated set
of cloud infrastructure components. Basically it is the synergy
of service-oriented and cloud architectures by means of a
consistent As-a-Service-Mechanism for basically infrastructure
cloud services. The fundamental characteristics of a Service-
oriented Infrastructure (SOI) are: business-driven infrastructure
on-demand, operational transparency, service measurement,
and consumer provider model. The SOCCI-Service-Oriented
Cloud Computing Framework is the extension of the Service-
oriented Infrastructure (SOI) mapped to the SOA Reference
Architecture [16]. The SOI-Framework is the layer on top of
the basic infrastructure and provides the elements of SOCCI:
Compute, Network, Storage, and Facilities. SOCCI extends
these basic elements of SOCCI by Business and Operational
SOCCI Management Building Blocks.
III. ENTERPRISE ARCHITECTURE ANALYTICS
The beneficial effects of data-driven decision making on the
performance of firms is well known [27]. So far, a data driven
approach has not been applied to enterprise analytics due to a
lack of information and limited computation capabilities.
However, nowadays the situation has changed due to the
advancement of virtualized infrastructure and big data.
Virtualized infrastructures such as cloud-environments [28] are
theoretically capable to describe every of the entities contained
in them. Furthermore, such environments not only have a static
perspective for architectural elements, they also contain a
dynamic perspective by logging all relevant elements in
virtualized environments. In the past, the data collected in
virtualized environments could not be exploited appropriately.
The huge sets of data overburdened the existing computation
capabilities. Furthermore, the data from the dynamic perspective
contains a lot of semi-structured data that was difficult to process
with existing approaches.
Through the advancement of Big Data, the situation has
changed significantly. Big Data [29] is one of the most
disruptive information technological developments [30] [31].
Big Data enables handling and analyzing more types of
unstructured (e.g. user statements in social media) and semi-
structured data as before [31] with higher velocity and volume
[32]. The importance of this disruptive development is shown by
an empirical study (worldwide online survey with over 1300 IT
managers) from ZDNet showing that "70% will use data
analytics by 2013" [33] (ZDnet 2012). Big Data is not identical
with a certain technology. However the use of the highly
distributed Hadoop [32] architecture is often associated with Big
Data. Big Data enables the extension of analytics in the three
dimensions volume, variety and velocity [34] as shown in Fig.
2. In comparison to business analytics it is now possible to
analyze nearly in real-time large quantities of data from data
sources with varying structure processed.
Fig. 2. Big Data scales out analytics based on [35]
Big Data applications are data-intensive applications, with a
large volume of data, a high velocity of processing and a data
variability of the existing IT solutions [31]. An example scenario
for Big Data is the provisioning of real-time information to
mobile users. Based on a stream of position information created
by the mobile user’s smartphone, the mobile receives
information selected from a variety of sources and provided
nearly in real-time. Various examples of real business cases
show business impact of Big Data such as [36]. In this case study
it is shown, that significant cost cuts could be achieved by
decreasing the estimated and actual arrival time of aircrafts.
Furthermore it is possible to increase sales through faster data
analysis and thereby better personalized promotions. Business
processes at the link to the customers and suppliers can be
improved by using Big Data [37]. This is possible because of the
increase of process quality, enabled by a better data quality for
decision making [37]. Furthermore, IT applications can benefit
Business
Analytics
Volume
Velocity
Variety
from the use of Big Data, because current empirical studies show
that the perceived advantages of Big Data are very high in the
field of technology development and IT [38]
Fig. 3. Enterprise Architecture Analytics
By combining conventional means of analytics such as data
warehouses [39] with Big Data it is possible to use both data
from the static and dynamic perspective of enterprise
architecture, as shown in Fig. 3. Especially structural
information from the static perspective may be processed using
a conventional extract, transform and load (ETL) [39] approach
in combination with a data-warehouse. The primary goal is to
gain descriptive information about the architecture. There is,
however, the possibility to use Big Data for tasks that are to
computation intensive.
The dynamics perspective of enterprise architecture provides
huge amounts of non- and semi-structured data from log files
etc. Using them predictive and even prescriptive analyses are
possible exploiting the capability of Big Data to process large
volumes of data. Furthermore, these analyses are available in a
shorter and shorter time scale. This and the decreased cost of
creating analyses allow the use Enterprise Architecture
Analytics not only in strategic decisions, but also tactical or even
operational decisions. Now, the analysis of Enterprise
Architectures using Business Analytics and Big Data shall be
described. The core idea is depicted in Fig. 3. Enterprise
Architecture is analysed using concepts and technologies
provided by Business Analytics and Big Data. By this mean
decision support is provided for designing and optimizing
Enterprise architecture.
The use of enterprise architecture analytics can generate
positive business value and business impacts. Important
architecture decisions can now be done based on a better data
quality and not only on a gut instinct. Therefore the following
metrics [40] [41] can be better acquired:
Cost metrics
Scalability metrics
Portability metrics
Security metrics
Etc.
A comparison of different architecture variants based on
metrics show CIO's a better view of the possibilities and
constraints of each. As a result better decisions can be applied.
This comparison can be applied for example by using a
utility analysis [42]. An example of a group utility analysis based
on the metrics above are shown in the following figure:
Fig. 4. : Example of utility analysis of architecture variants
Furthermore a better understanding of processes in the
enterprise architecture based on a larger base of data can
improve IT services. If the demand of each IT service at each
time is known or better predicted, preproduction costs can be
decreased. Prediction algorithms and methods for the calculation
of the demand of IT services (e.g. ARIMA [43], linear regression
[44], neuronal networks [45] models) and their forecasts of
demand be improved by a broader database. Generally the IT-
Demand (e.g. normal use and peaks) are difficult to predict [46].
The prediction can be improved, because of using a broader
database and using complex as well as robust (e.g. non-linear
analysis) prediction algorithms (e.g. linear regression [44]) [47].
Therefore forecast errors can be decreased. A Forecast error is
defined as the absolute value of the difference between the real
and predicted demand [48] [49]:
Equation 1: forecast error
Semi-
structured
data
Non-
structured
Data
Structured
Data
ETL
Architecture
Analytics
Data
Warehouse
Big Data
Architecture
Decision
Support
Static
Perspective
Dynamic
Perspective
Strategic
Tactic
Operational
Integration Storage Analysis Decision
Descriptive
Predictive
Prescriptive
Enterprise Architecture
The quality of prediction the forecast error can be improved by
using other more detailed metrics (like RMSE [48] [49] ).
A reduced forecast error outcome a cost saving for idle time
costs:
Equation 2: forecast error
Idle time is defined as the time where not all available IT
resources are used. Furthermore some effort for preproduction is
needed, because many IT services need some time or capacity to
be established.
Therefore new price differentiation (internal or external) can
be implemented. Times of less demand can decrease with a cost
reduction of the lower price limit for the internal or external
customers. In contrast times of a huge amount of demand can be
managed be a increased price for customers. As a result idle time
costs can be decreased and earnings increased. These facts are
shown in the following formula.
Equation 3: cost aspects of IT services
The budget variable in equation 3 describes the budget of the
IT department for enterprise architecture solutions. The IT budget
includes the costs of salaries, maintenance, technology, R&D etc.
[50]. Mostly this budget is defined by the CFO of each enterprise
or organization. Each used capacity cost (e.g. cost for database
access, business process use, esb-use, etc) are defined in the
used_capacity_cost variable. Furthermore other costs (e.g. for
administration, special indirect costs) defined by other_costs
variable in equation 3.
An increase of incomes will implement trough a reduction of
idle time costs through a better knowledge about the demand. If
the IT department is implemented as a profit center, earnings will
increase and improve the standing of the department in the
enterprise or organization. In this case the variable budget must be
renamed as revenue in the formula.
In the past IT departments were distinguished as a cost driver
and collection of risks [51]. An applied enterprise architecture
analytics approach can improve the standing of the IT department
in the enterprise and maybe move the IT department from a cost
driver to a profit center. This is possible because of more accurate
enterprise architecture decisions and focusing on cost aspects,
which are better supported by this approach than in the past and
increase the outcomes of the enterprise (e.g., productivity [50].
IV. EA CAPABILITY ASSESSMENT
From an EA capability perspective, the core question in
context of analytics and big data is which capabilities significantly
would benefit from big data analytics. This section will investigate
this question and derive requirements to big data analytics (BDA).
Starting point for the discussion will be the capability catalog as
part of an EAM maturity model developed by Wißotzki, which is
published in [52] [53] .
Maturity models are specific management instruments, which
define various degrees of maturities in order to evaluate to what
extent a particular competency fulfils the qualitative requirements
that are defined for a set of competency objects [54] and the
development processes in organisations [55]. This abstract set of
competency objects, which are representations of concrete objects
from the real world. Beginning with very early stages of these
entities, maturity models define anticipated, logical and
consecutive development paths until observed objects reach an
absolute maturity [56]. Having their origins in the software
industry, maturity models are designed to measure the current state
- the achieved level of competence - by means of assessment
methods [57] [58].
The idea of EAM paradigm is modelling of the important
enterprise elements and their relationships, which allows the
analysis of as-is and target state dependencies [59]. In this context,
EA models serve as maps with information of the current situation
and strategies for future directions of the company [60]. Making
the organizations more sensitive towards the interaction of
business strategies, customers, application systems and
organizational units, companies need to control enterprise-wide
processes and adopt matching actions [61]. For this purposes, the
concept of maturity was employed for EA which assigns different
levels of achievement by means of a maturity assessment to
processes, sub-processes, capabilities and characteristics [57].
Organizations will increasingly adopt maturity models to guide
the development and implementation of their strategies. Yet it is a
challenge for an organization to efficiently put the right
capabilities into practice. In order to do so, organizations have to
carry appropriate actions into execution, which later on should be
turned into so-called “initiatives”. For these actions to be taken
there is a need for an integrated approach, which could be gained
by implementing EAM. This is a prerequisite for an enhanced
holistic enterprise view that reduces the management complexity
of business objects, processes, strategies, information
infrastructure and the relations between them. Nevertheless the
successful adoption of EAM is accompanied by challenges that an
![Fig. 2. Big Data scales out analytics based on [35]](/figures/fig-2-big-data-scales-out-analytics-based-on-35-3c8v1ymo.webp)
