An Approach for Transforming IT-Network Diagrams into
Enterprise Architecture Models
Daniel Maciejewski and Nicolas Mayer
Luxembourg Institute of Science and Technology, Esch-sur-Alzette, Luxembourg
maciejewski.dani@gmail.com, nicolas.mayer@list.lu
Keywords: Network Diagram, Enterprise Architecture Model, ArchiMate, Model Transformation, CISCO.
Abstract: Enterprise Architecture (EA) models propose to capture the activities of an organization, going from its
business aspects to its IT infrastructure. Such an approach is promising to support reasoning on specific
concerns, especially those relying on the business-to-IT stack, such as enterprise transformation, security, IT
investment, etc. However, most of the organizations do not have existing EA models, and are reluctant to
establish them, especially from scratch, mainly due to the length and complexity to do so. Our insight is that
we can leverage on network diagrams, one of the most common kind of models available in organizations, to
generate part of EA models. In this paper, we propose an approach to transform network diagrams into EA
models. In this context, we focus on CISCO as the reference for network concepts, and ArchiMate as the
standard modelling language for EA.
1 INTRODUCTION
Enterprise Architecture Management (EAM) has
shown to be a valuable and engaging instrument to
face enterprise complexity and the necessary
enterprise transformation (Saha, 2013; Zachman,
1987). It offers means to govern enterprises and make
informed decisions: description of an existing
situation, investigation and expression of strategic
direction, analysis of gaps, planning at the tactical and
operational level, selection of solutions, and
architecture design (Op’t Land et al., 2008). As part
of a global EAM adoption, describing Enterprise
Architecture (EA) with a suited language (i.e. EA
modelling) is considered as a key activity (Lankhorst,
2005) and the scope of our paper is focused on this
concern.
However, the current problem is that most
organisations pay few attention to the modelling of
their structure. Among the current limitations to a
broader adoption of EA modelling, we particularly
noticed (Lankhorst, 2005):
EA modelling is a complex task and requires
specific skills
EA modelling is a time consuming task, especially
when started from scratch
There are numerous and disregarded modelling
tools that can be used
To deal with these issues, our main assumption is
that the reuse of existing material (i.e. existing
models) would help in the development and adoption
of EA models and also reduce their development
length. In this context, IT network diagrams are usual
existing material in the structure of an organization,
and appear to be a good basis to EA modelling.
Indeed, each structure needs network models in order
to support infrastructure design and most of them
have them available.
In this paper, our aim is to propose an approach to
use IT network diagrams to generate (part of) EA
models. Our insight is that it is a relevant and realistic
beginning to develop EA models, allowing saving
time by accelerating part of the EA modelling tasks.
It is worth to note we consider that such an approach
would never produce complete and satisfactory EA
models, but we see it as an interesting trigger to start
the design of such models. In order to limit the scope
of this project, we will focus on ArchiMate (The
Open Group, 2013) as the exploited EA Modelling
Language (EAML), but we are aware that others
could be relevant too (BPMN (Object Management
Group, 2011), UML (Fowler, 2003), etc.).
This paper is structured as follows. In the
subsequent section, we provide some background
knowledge about the main literature we use: CISCO
as a de facto standard for Network Architecture
Design and the ArchiMate Language, especially its
186
technology layer. Section 3 describes the first step of
our research method that is the definition of a network
concept classification, followed, in Section 4, by the
integration of these concepts into ArchiMate. We
finish with concluding remarks and future work.
2 BACKGROUND
2.1 CISCO as a de facto Standard for
Network Architecture Design
Network architecture design is the science to design
good networks, making them safe and available
(Stewart et al., 2008). IT Network concepts are
numerous, especially because of the number of
different technologies, products, companies, etc. We
found a lot of information in uncertified articles or
unprofessional tutorial to design a network diagram.
However, to the best of our knowledge, there is a lack
of standard or reference model depicting the concepts
at stake, and in our context, it is a major issue. The
current state-of-practice is actually the use of
technology-specific terms and concepts. By
reviewing major modelling software proposing
network architecture modelling features (VISIO,
Gliffy…), our conclusion is that CISCO is considered
as a de facto standard. Hence, we decided to focus on
CISCO concepts, expecting to cover a major
proportion of existing concepts. The survey of
literature for network architecture is consequently
focused on CISCO. We distinguish three main
sources of reference material related to CISCO:
CISCO documentation: a set of documents
listing main CISCO network concepts (e.g.
Technology, protocol, or product). In this
category, we identified the following documents:
CISCO Iconography (Cisco Systems Inc., 2014),
CISCO Product Quick Reference Guide (Cisco
Systems Inc., 2013).
CISCO Website: the CISCO website is a field-
based source. It is the first interface between
CISCO up-to-date products and clients.
Modelling Software: most of modelling software
uses modelling set of objects called “stencils”, one
among the available ones being generally based
on CISCO concepts. We have only used those for
which the CISCO stencil was sufficiently
developed and specified, and we have set apart the
others (e.g. Microsoft VISIO). The analysed
modelling software are:
o Gliffy (Gliffy, 2017): Gliffy is a cloud-based
diagramming web application founded in
2005, allowing collaborative work. In our
work, we focus specifically on the library
concerning CISCO objects.
o CISCO Packet Tracer (Cisco Systems Inc.,
2017): Packet Tracer is a simulation program
designed by CISCO Systems. The software
allows users to create network topologies and
reproduce nowadays computer networks.
2.2 ArchiMate Technological Layer
Enterprise Architecture (EA) is defined as a coherent
whole of principles, methods, and models that are
used in the design and realisation of an enterprise’s
organisational structure, business processes,
information systems, and infrastructure (Lankhorst,
2005). To provide a uniform representation for
diagrams that describe EA, the ArchiMate modelling
language (The Open Group, 2013) has been produced
by The Open Group, an industry consortium
developing standards. It offers an integrated
architectural approach to describe and visualize the
different architecture domains and their underlying
relations and dependencies. The role of the
ArchiMate standard is to provide a graphical
language for the representation of EA over time, as
well as their motivation and rationale. It is today a
widely accepted open standard for modelling EA
(Vernadat, 2014), with a large user base and a variety
of modelling tools that support it.
Figure 1: Definitions and visual representation of concepts
of the ArchiMate technology layer.
ArchiMate proposes a 3-layered architecture for
structuring its core concepts: the Business Layer,
Application Layer, and Technology Layer. In our
perspective focused on the modelling of network
An Approach for Transforming IT-Network Diagrams into Enterprise Architecture Models
187
infrastructure, we will particularly analyse the
Technology Layer that provides infrastructural
services needed to support business. Figure 1,
extracted from the ArchiMate 2.1 Specifications (The
Open Group, 2013), gives an overview of the
Technology Layer concepts, including definition and
visual representation for each concept.
3 ALIGNMENT AND
INTEGRATION BETWEEN
CISCO AND NETWORK
CONCEPTS
Alignment between CISCO and network concepts is
the first phase of our contribution. First, for each
analysed literature reference, we need to determine
concept classes (i.e. a concept classification for each
literature reference). Each class can be seen as a set
of modelling objects (or instances of generic
concepts) and the set of classes creates a taxonomy of
network concepts of a given literature reference.
Then, we need to map the defined taxonomies
together thanks to a classification methodology. The
objective of this alignment is to propose a
consolidated list of network generic classes. The
classification methodology uses refinement cycles
based on criteria to do the most relevant arrangement
between network classes. Thus, each chosen
classification pass through each criterion cycle, which
correspond to different levels of details and analysis:
Syntax: the first level is the semantic analysis of
the class i.e. name’s similarities or spelling
syntax. For example, a grouping called “Routing”,
can easily be associated to another grouping
called “Routers”.
Network subdomains: the second level of
analysis aims at determining, for each instance of
the class, to which network fields it belongs. For
example, “Connecting Safety and Security” gather
physical security concepts (doors, cameras, etc.)
and IT security concepts (firewall, guards, etc.). It
helps us to exclude concepts which are out of our
scope (in that case Physical Security concepts).
Functionality: To remove any ambiguity, a
detailed analysis of each instance functionalities
is required. In fact, an instance can belong at a
time to different classes (or to different network
subdomains). There is a need to analyse
objectively to which class each instance belongs
the most. For example, a wireless router can
belong to the “Wireless class” and to the “Router
class”. Thanks to the predominant functionality,
we can say that a wireless router is a router and
effectively belong to the “Router class”.
We mainly focus on the most relevant sources of our
survey, in order to make these “instance’s groups” the
most significant towards the network engineering
field. First, we focus on the CISCO Documentation
(Cisco Systems Inc., 2014, 2013), also including the
CISCO website. Then, we chose Gliffy Online
Platform (Gliffy, 2017), and finally, the simulation
software CISCO Packet Tracer (Cisco Systems Inc.,
2017). Then, we applied the classification
methodology described previous1y on each
reference, to refine the network classification. The
output of the process creates a classification for our
instances. We’re getting the so-called “Network
concepts classification”, presented below.
This consolidated list expresses main CISCO-
based generic network concepts. By using this
classification, we will be able for instance to classify
other concept’s instances coming from other brand
models. We use a granularity generic enough, which
allows us to have a viable classification on time. We
give a description for each class of our network
concepts classification and examples of what we find
in each of them:
Collaboration: It regroups all instances that aim with
the teamwork. We find conferencing devices, IP
phones, communicating software, etc.
Security: This class regroups all security products
including physical security and IT security. Thus, we
have devices such as monitoring cameras, firewalls,
etc. Also, we can find some security software
products (e.g. IOS Firewall).
Switch: The class gather all different types of
switches that exist in some various forms depending
on the technology used or the functionalities
incorporated (e.g. Multilayer Switch, etc.).
Router: The class regroups the various types of
routers (e.g. Router with Firewall, NetFlow Router,
etc.).
Wireless: Collection of wireless devices, or wireless
modules (e.g. access points, WLAN controller, etc.)
Software: This is a class which gather different type
of software solutions, mainly from CISCO. We
especially find system software, like OS.
Transport/Telephony: This class gather all devices
that concern the physical devices of telephony and
transport. We find mainly devices that deals with the
first layer of the OSI model like modems, DSLAM,
etc.
End Device: It concerns end devices like personal
computers, TVs, printers, etc.
Seventh International Symposium on Business Modeling and Software Design
188
Physical Server: It regroups the different types of
existing servers like Web servers, certificates servers,
file servers, etc.
Data Storage: This class is derived from the physical
servers’ class. Indeed, because of the huge number of
existing data storage types’ instances (storage
module, data monitoring software, etc.) it is essential
to distinguish it from the physical devices.
Management: It is a field’s class that gathers
management devices, modules and software. In big
networks, it is crucial to consider management
devices, or management software to access, control,
or monitor devices.
Hub: Similarly, to the Router class or the Switch
class, it regroups hub concepts and instances. This
type of devices is tending to disappear because of the
technological advancement. However, we have to
take into account old networks that could still have
this type of devices.
Protocol: This class is particular because in general,
protocols cannot be depicted in an architecture, apart
from writing the name besides a device or a
connections. However, CISCO has created protocols’
concepts like IP protocols, or FDDI Ring that allow
representing some kind of protocols.
Topology: Such as the protocol class, it is possible to
represent the topology of a network.
Connections: In a network diagram, connections are
represented in different manners. For example, in
CISCO Packet Tracer, we are able to recognize
optical link, wireless communication, etc. This class
collects all type of connections that can be
materialized in a network diagram.
Network Architecture: It represents different types
of network architectures like Cloud, CISCO layered
architecture, etc. In diagrams that use modularization,
it is possible to represent a network architecture
through a dedicated icon representing this concept.
4 CONSTRUCTION OF A
NETWORK-ARCHIMATE
METAMODEL
4.1 Conceptual Alignment
After having analysed CISCO concepts with their
instances, we have created the network concepts
classification following our methodology, as depicted
in the previous section. As a reminder, the goal of our
research work is to use network diagrams so as to
generate a basis of EA models. Therefore, this second
step aims at creating a relationship between our
network concepts classification and ArchiMate
Technology Layer concepts. Thus, in order to
integrate these global concepts (i.e. classes) into the
ArchiMate metamodel, we need to respect the
ArchiMate structure (i.e. rules and concepts) and use
the ArchiMate metamodel as an input. We called this
relationship the “integration step”.
In order to do this, it is necessary to add network
generic classes to the ArchiMate Technology Layer
metamodel. Consequently, we will be able to
represent a network diagram with the ArchiMate
Technology Layer concepts, thanks to the metamodel
rules (i.e. links between concepts of the metamodel).
To allow this conceptual alignment, we use
relationships inspired from the approach of semantic
correspondence between concepts from Zivkovic et
al. (Zivkovic et al., 2007). Themselves inspired by
UML relationships (Fowler, 2003), we explain these
relationships below:
Equivalence: concept A is semantically
equivalent to concept B;
Generalisation: concept A is a generalisation of
concept B, i.e. concept B is a specific class of
concept A;
Specialisation: concept A is a specialisation of
concept B, i.e. concept B is a generic class of
concept A;
Aggregation: concept A is composed of concept
B, i.e. concept B is a part of concept A;
Composition: concept A is composed of concept
B (with strong ownership), i.e. concept B is a part
of concept A and does only exist as part of concept
A;
Association:
concept A is linked to concept B
By confronting each instance (of network generic
classes) with ArchiMate, we realize that we can affect
to each instance one of the Active Structure concepts
of the ArchiMate Technology Layer, i.e. Node,
System software, Device, Communication path,
Infrastructure interface and Network. Actually, the
ArchiMate framework can represent its own network
diagrams, as CISCO, that’s the reason there are some
similarities between our created network generic
classes and Active Structure concepts. Table 1 show
these equivalences we identified. Table 1 presents
two types of equivalences:
First, we have equivalences between some
instances of (network generic classes) and leaf’s
concepts of the ArchiMate metamodel (Device,
Node, System Software). These latter are the most
used concepts for network diagram conception in
the EA context.
An Approach for Transforming IT-Network Diagrams into Enterprise Architecture Models
189
Secondly, we noticed that some classes can be
mapped as is to ArchiMate main concepts. For
example, instances of our Connections class are in
some way equivalent to the ArchiMate
Communication Path class instances.
Table 1: Equivalence between type of instances and
ArchiMate concepts.
To use Network generic classes as concepts of
ArchiMate, we have to add our concepts to the
ArchiMate metamodel without modifying its
structure. In fact, we can only improve the metamodel
by specializing ArchiMate concepts or finding
equivalence between both types of concepts. Thus,
there is two relevant relations from Zivkovic et al. we
can use: specializations and equivalences.
Specialization is a type of semantic mapping used
for concepts that are a part of ArchiMate concepts
(see Figure 1). For example, physical router devices
of our Router class are specializations of the
ArchiMate Device concept. Thus, there is a need to
split our network generic classes in order to make
possible the addition of instances to the ArchiMate
metamodel, and thus to be compliant with ArchiMate.
Table 2 expresses the different partitions we had
to make in order to be able to specialize the
ArchiMate metamodel with our concepts. Table 2
presents three main type of classes:
Network generic classes that are enough generic
to propose Device instances, System Software
instances and Module instances, which are the
most used ArchiMate concepts for network
diagramming in EA context. We called these
classes “Field” because of their huge number of
different type of instances.
Some network generic classes are specifically
concerning hardware instances. In the ArchiMate
context, it concerns only physical type of
instances (Physical router, End devices, etc.)
Remaining network classes can’t be divided,
because of too specific instances, such as specific
topologies, or protocol. In that context, it is
irrelevant to subdivide these network generic
classes. We called them the “Support” classes.
Table 2: Partition of network generic classes.
Then, we create a detailed alignment table
between previously subdivided network generic
classes and ArchiMate concepts. Table 3 presents the
semantic mapping:
Table 3: Detailed alignment table.
Subdivided Network
Generic Classes
ArchiMate
Semantic
mapping
type
Collaboration Device Device
Specialization
Collaboration System
Software
System Software
Collaboration Module Module
Security Device Device
Security Module Module
Security System
Software
System Software
Data Storage Device Device
Data Storage Module Module
Data Storage System
Software
System Software
Transport/Telephony
Device
Device
Transport/Telephony
System Software
System Software
Transport/Telephony
Module
Module
Management Device Device
Seventh International Symposium on Business Modeling and Software Design
190
Table 3: Detailed alignment table. (cont.)
Subdivided Network
Generic Classes
ArchiMate
Semantic
mapping
type
Management System
Software
System Software
Specialization
Management Module Module
Switches Device Device
Switches Module Module
Wireless Device Device
Wireless Module Module
Routers Device Device
Routers Module Module
Physical Servers Device Device
Physical Servers
Module
Module
End Devices Device Device
End Devices Module Module
Hub Device Device
Hub Module Module
Protocol Interface
Topolog
y
Network
Network Architecture Network
Connections
Communication
Path
Equi.
4.2 Proposed Network-ArchiMate
Metamodel
Based on the alignment performed, we need to inte-
grate the classes of our network classification into the
existing ArchiMate metamodel, following and
complying with the existing structure and associated
rules. The mapping table permits us to integrate the
concepts into the ArchiMate metamodel. Thanks to
the semantic rules, we are able to propose the
integrated Network-ArchiMate metamodel, as
represented in Figure 2. We give also the example of
Firewall instances, which we scattered into
corresponding type of concept (i.e. Device, System
Software, Module).
5 CONCLUDING REMARKS
The approach presented in this paper aims at
proposing a way to transform network diagrams into
EA models. First, we had to suggest a network
concepts taxonomy that allows us to classify each
network concept and its instances in these generic
classes. Then, by aligning these classes with
ArchiMate concepts, we proposed a metamodel
which allows us to integrate network-specific
concepts with ArchiMate.
The approach presented here, and especially the
integrated metamodel, have been used in a fictitious
case to experiment the transformation from network
diagram to an ArchiMate model. This experiment, not
described here for sake of brevity, has shown that our
Figure 2: Proposed Network-ArchiMate metamodel.
An Approach for Transforming IT-Network Diagrams into Enterprise Architecture Models
191
approach is applicable. It has also shown some ways
of improvement, such as the introduction of
additional specifications (e.g. a color code) in the
ArchiMate language in order to keep trace of the
original network class, this information being lost in
the transformation proposed. However, no conclusion
can be drawn from this experiment at the level of
soundness and usefulness of our approach in a real-
world context and further validation work is deemed
as necessary to consider our approach as valid.
Regarding future work, we first need to
experiment our approach on a wider and real world
case. Then, it is necessary to assess with users the
relevance of the approach especially at the level of the
soundness of the EA models obtained compared to
the initial network models, as well as their usefulness
as a starting point to design complete EA models.
ACKNOWLEDGEMENTS
Supported by the National Research Fund,
Luxembourg, and financed by the ENTRI project
(C14/IS/8329158).
REFERENCES
Cisco Systems Inc., 2013. Cisco Product Quick Reference
Guide.
Cisco Systems Inc., 2014. Cisco Iconography.
Cisco Systems Inc., 2017. Packet Tracer.
Fowler, M., 2003. UML Distilled: A Brief Guide to the
Standard Object Modeling Language. Addison-Wesley
Longman Publishing Co., Inc., Boston, MA, USA.
Gliffy, 2017. Gliffy Online User Manual.
Lankhorst, M. (Ed.), 2005. Enterprise Architecture at
Work: Modelling, Communication And Analysis.
Springer.
Object Management Group, 2011. Business Process Model
and Notation (BPMN) 2.0.
Op’t Land, M., Proper, H.A., Waage, M., Cloo, J., Steghuis,
C., 2008. Enterprise Architecture - Creating Value by
Informed Governance, Enterprise Engineering.
Springer Berlin Heidelberg.
Saha, P. (Ed.), 2013. A Systemic Perspective to Managing
Complexity with Enterprise Architecture: IGI Global.
Stewart, K., Adams, A., Reid, A., Lorenz, J., 2008.
Designing and Supporting Computer Networks, CCNA
Discovery Learning Guide, 1st ed. Cisco Press,
Indianapolis, Ind.
The Open Group, 2013. ArchiMate® 2.1 Specification (
No. Open Group Standard (C13L)).
Vernadat, F., 2014. Enterprise Modeling in the context of
Enterprise Engineering: State of the art and outlook.
International Journal of Production Management and
Engineering 2, 57.
Zachman, J.A., 1987. A framework for information systems
architecture. IBM Systems Journal 26, 276–292.
Zivkovic, S., Kuhn, H., Karagiannis, D., 2007. Facilitate
Modelling Using Method Integration: An Approach
Using Mappings and Integration Rules. In: Proceedings
of the 15th European Conference on Information
Systems (ECIS 2007).
Seventh International Symposium on Business Modeling and Software Design
192