How to Measure Enterprise Architecture Complexity: A Generic Approach, Practical Applications, and Lessons Learned The Open Group Conference – London 2013 Dr. Christian Schmidt London, October 21st, 2013
How to Measure Enterprise Architecture Complexity:
A Generic Approach, Practical Applications, and Lessons Learned
The Open Group Conference – London 2013
Dr. Christian Schmidt
London, October 21st, 2013
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 2
How to deal with complexity?
Questions to be answered
What does complexity
mean to the enterprise
architect?
How can complexity be
measured?
Who is interested in
complexity analyses?
What are the possible usage
scenarios for complexity
measures?
Which practical
experiences have
been made?
What are the lessons
learned?
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 3
Research cooperation
Research projects that we are currently involved in
CEAR: Complexity in Enterprise
Architectures
CALM3: Complexity of Application Land-
scapes – Models, Measures, Management
how can complexity be defined and
quantified in the context of enterprise
architectures?
what impact does complexity have on
the enterprise success?
which principles should be followed in
the management of complexity
what does complexity mean in the
context of application landscapes?
what are the drivers for complex
application landscapes?
how can the complexity of application
landscapes be measured?
how can complexity models be used
for application landscape planning?
Project 1: (started early 2012) Project 2: (started early 2013)
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 4
Agenda
1. The Role of Complexity
2. Measuring EA Complexity
3. Case Example: Application Landscape Planning
4. Lessons Learned and Recommendations
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 5
Facts on enterprise architecture complexity
Some observations from enterprise reality
A large industry firm has taken records of over 4.000 business applications
A large commercial bank employs more than 25 different DBMS versions
A universal bank has not turned off batches for years due to unknown side effects
Contrary to the IT strategy, the board of an insurance firm was equipped with iPads
The capital markets division of a public sector bank developed its own data warehouse
During an audit, a reinsurance group identified 50 previously unknown EUC applications
The EA tool of an insurance firm reached three times the expected size after only one year
During a transformation project, the number of known as-is applications more than doubled
Two architects of a pharma group, who attended a conference, had never heard of each other
Complexity
is everywhere
Complexity
is created
constantly
Complexity
is generally
underestimated
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 6
The law of rising complexity
Complexity grows over time if no specific action is taken
According to the Second Law of Software Evolution, the complexity of a software system in use will
increase with time if no explicit action is taken to avoid this [Le97]. It may be argued that the law will also
hold on a macro level (as an aggregation of the many individual evolution processes) [SB11].
Required
Business
Complexity
Business
Complexity
Surplus
Required IT
Landscape
Complexity
IT Landscape
Complexity
Surplus
Time
Complexity
see also [Ru13]
Drivers
• Mergers &
Acquisitions
• Local Optimization /
Politics
• Mergers &
Acquisitions
• Local Optimization /
Politics
• Development of
Technology /
IT Industry
• Business Model
• Legal Regulation /
Supervision
Constituents
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 7
The central role of complexity
Drivers and outcomes of IT complexity
IT
Complexity
1. Local optimization of
business departments
2. Rising business
complexity
3. Increasing legal
regulation
4. Technological
progress
5. Short-term optimization
of business departments
1. Increase of
communication efforts
2. Increase of
development costs
3. Increase of
maintenance costs
4. Increase of
implementation periods
5. Increase of
start-up times
The TOP 5 drivers and outcomes of IT complexity according to a survey within the CALM3 working
group [Sc13].
CALM³
Complexity of Application
Landscapes – Models,
Measures, Management
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 8
The role of the enterprise architect
Active management of the complexity surplus
Due to the negative impacts of complexity, architectures should only be as complex as required.
It is a core responsibility of the architect to manage the complexity surplus.
Required
Business
Complexity
Business
Complexity
Surplus
Required IT
Landscape
Complexity
IT Landscape
Complexity
Surplus
Time
Instruments Constituents
However, non-functional requirements (like vendor dependency, security) and specific context factors
(like the business model or sourcing strategy) need to be taken into account as well.
Complexity
• Shared Service
Centers /
Centralization
• Standard Processes
• Sourcing
• etc.
• Target Landscapes
• Books of Standards
• Reference
Architectures
• Governance
Processes
• etc.
Bu
sin
es
s A
rch
ite
ct
IT A
rch
ite
ct
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 9
The complexity continuum
The right amount of complexity in the right place
Minimal complexity is often neither required nor appropriate. Instead, the optimal level of complexity
may vary across architecture layers and domains.
Technical
Architecture
Application
Architecture
Business
Architecture
lower operating
costs
lower maintenance
costs
lower procurement
costs
higher agility /
reduced change
efforts
etc.
higher customer
value
higher business
value
higher
robustness
reduced vendor
dependency
etc.
For differentiating front-end domains, for instance, a higher level of complexity may be appropriate
than for non-differentiating back-end domains.
TD1 TD2 TD3 TD4
AD1 AD2 AD3 AD4
BD1 BD2 BD3 BD4
low complexity high complexity
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 10
Agenda
1. The Role of Complexity
2. Measuring EA Complexity
3. Case Example: Application Landscape Planning
4. Lessons Learned and Recommendations
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 11
The different faces of complexity
Stakeholders and their view of complexity
Head of Solution
Development
• Number of Services
• Number of Vendor Relations
• Number of Applications
• Number of Development Lines
• Number of Interfaces
• Number of Products
• Number of Process Variants
• Number of Frameworks
• Lines of Code
• Number of Applications
(within Domain)
• Number of Interfaces
(within Domain) • Number of Platforms
• Number of Dependencies
Chief Information
Officer (CIO)
Line of Business
Manager
Software
Developer
Domain
Architect
Head of
Operations
• Applications per Capability
• Number of Platforms
• Number of Interfaces
Enterprise
Architect
Complexity is a fuzzy term. Different stakeholders usually have different views and conceptions.
illustration
exemplary
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 12
Traditional ways to measure complexity
Use of specific complexity indicators
How to ensure completeness?
How to compare different
indicator values?
How to aggregate indicator
data?
What is the information value
of average numbers?
How can the required data be
collected?
Existing approaches to measure EA complexity are mostly based on a set of specific complexity
indicators. This has a number of shortcomings:
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 13
Generic approach proposed by CEAR
Common definition of structural complexity
Number
(N)
Heterogeneity
(H)
Elements
(T)
Relationships
(R)
NT NR
HT HR
Business Architecture
Application Architecture
Technical Architecture
The structural complexity of an Enterprise Architecture (interpreted as a system) may hence be
defined as a tuple CEA = (NT, HT, NR, HR).
(43 cases) (24 cases)
(31 cases) (10 cases)
According to a literature review (47 relevant publications out of 4.000 articles in international IS journals
and conferences), most authors associate system complexity with the following aspects [SWK13]:
see also [SM03]
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 14
Quantifying heterogeneity
Heterogeneity as a statistical property
Definition:
Heterogeneity is defined as the diversity
of elements or relationships of a system
with respect to certain characteristics
(attribute values).
0%
10%
20%
30%
40%
50%
60%
DB/2 MS SQL Oracle Sybase
Sh
are
Based on this, well-studied concentration
measures may be adopted from other research
domains like economics / monopoly regulation
(e. g., Herfindahl-Hirschman-Index) or biology
(e. g., Simpson-Index) [Wi12].
Heterogeneity can be captured as a statistical
property and be described by means of empirical
frequency distributions.
Following the above approach, the problem of
quantifying complexity is reduced to quantifying
heterogeneity.
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 15
Quantifying heterogeneity
Measure requirements
Requirement Scenario A Scenario B Expected Result
Impact of Small
Shares
Shift in the
Classification
No Impact of
Proportional Changes
Additional
Characteristics Het(A) < Het(B)
Het(A) < Het(B)
Het(A) < Het(B)
Het(A) = Het(B)
0%
50%
100%
MS SQL
Oracle0%
50%
100%
MS SQL
Oracle
DB/2
0%
50%
100%
MS SQL
Oracle0%
50%
100%
MS SQL
Oracle
0%
50%
100%
MS SQL
Oracle0%
50%
100%
MS SQL
Oracle
DB/2
500
150
0
500
1000
1500
MS SQL
Oracle
1000
300
0
500
1000
1500
MS SQL
Oracle
[SWK13]
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 16
Quantifying heterogeneity
The entropy measure
0%
10%
20%
30%
40%
50%
60%
DB/2 MSSQL
Oracle Sybase
0%
10%
20%
30%
40%
50%
60%
DB/2 MSSQL
Oracle Sybase
𝐸𝑀 = 𝑓𝑖 𝑙𝑛1
𝑓𝑖
𝑛
𝑖=1
𝑤𝑖𝑡ℎ 𝑓𝑖 = 𝑥𝑖 𝑥𝑗𝑛𝑗=1
The requirements are best met by the Entropy
Measure as introduced by Shannon [Sh48]
[JB79] [SWK13]:
Interpretation of the Entropy Measure is facili-
tated by the so-called Numbers Equivalent
Entropy Measure, which denotes the equi-
valent number of characteristics at equal
distribution:
𝐸𝑀𝐴 = 𝑒𝐸𝑀
𝐸𝑀 = 1.12
𝐸𝑀 = 1.10
𝐸𝑀𝐴 = 𝑒1.12 = 3.06
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 17
Operationalization
Model application based on EA repository data
EA
Repository
In order to apply the model, a further differentiation of element and relationship types is necessary.
Beyond that, data is required in an appropriate form.
Architecture repositories with their data structured along EA metamodels provide the ideal basis
for data collection and the calculation of complexity measures [SWS13].
EA Metamodel
source: [TOG13]
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 18
Complexity aspects
Types of complexity aspects in metamodels
• Attr. 1
• Attr. 2
• …
ET 1
• Attr. 1
• Attr. 2
• …
ET 1
• Attr. 1
• Attr. 2
• …
ET 2
• Attr. 1
• Attr. 2
• …
ET 1
• Attr. 1
• Attr. 2
• …
ET …
• Attr. 1
• Attr. 2
• …
ET n
Type 1:
Element Types
Type 2:
Relation Types
(Special Case of Type 3)
Number of
element
instances of a
certain type
Number of
relation
instances of a
certain type
Number of
path
instances of a
certain type
Heterogeneity of
element instances of a
certain type with respect
to a defined characteristic
(attribute of the type)
Heterogeneity of relation
instances of a certain
type with respect to a
defined characteristic of
one of the element types
involved
Heterogeneity of path
instances of a certain
type with respect to a
defined characteristic of
one of the element types
involved
[SWS13]
Type 3:
Path Types
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 19
Example: Complexity aspect of type 1
Applications by vendor
SAP FI
XIA
Elena
ResQ
0%
20%
40%
60%
80%
100%
SAP SD TW CG
0%
20%
40%
60%
80%
100%
SAP SD TW
N = 4 H = 1.39 N = 4 H = 1.04
Sta
nd
ard
iza
tion
SAP FI
XIA
SAP FS-CD
ResQ
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 20
Example: Complexity aspect of type 1
Applications by vendor
SAP FI
XIA
Elena
ResQ
0%
20%
40%
60%
80%
100%
SAP SD TW CG
0%
20%
40%
60%
80%
100%
SAP TW CG
N = 4 H = 1.39 N = 3 H = 1.10
De
co
mm
issio
nin
g
SAP FI Elena
ResQ
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 21
Example: Complexity aspect of type 2/3
Application usage by applications
SAP FI
XIA
UK Branch
DK Branch
SAP FI
XIA
UK Branch
DK Branch
0%
20%
40%
60%
80%
100%
SAP FI XIA
0%
20%
40%
60%
80%
100%
SAP FI XIA
N = 4 H = 0.69 N = 2 H = 0.69
Dis
en
tan
gle
me
nt
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 22
Example: Complexity aspect of type 2/3
Application usage by applications
SAP FI
XIA
UK Branch
DK Branch
SAP FI
XIA
UK Branch
DK Branch
0%
20%
40%
60%
80%
100%
SAP FI XIA
0%
20%
40%
60%
80%
100%
SAP FI XIA
N = 2 H = 0.69 N = 2 H = 0
Co
ns
olid
atio
n
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 23
Agenda
1. The Role of Complexity
2. Measuring EA Complexity
3. Case Example: Application Landscape Planning
4. Lessons Learned and Recommendations
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 24
Case example
Application landscape planning
[SWS13]
Project Outline
Mission: Develop Target Application Landscape
Industry: Special Insurance
Period: May – Dec 2012
Drivers: high redundancy, poor acceptance of core
systems, extensive use of EUC components
(Access, Excel, etc.)
Objectives: elimination of EUC applications, regional
consolidation, technological standardization
and alignment with group standards,
introduction of a centralized integration platform
Methodology: TOGAF ADM / ADD [TOG13]
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 25
Case example
Metamodel of the organization
Line of Business Business
Function Organization Unit
Application
Domain
Application
Component
Platform
Infrastructure
System
Technical
Domain
Information Entity
Application
Interface
Ap
pli
cati
on
Arc
hit
ectu
re
Bu
sin
es
s
Arc
hit
ectu
re
Te
ch
nic
al
Arc
hit
ectu
re
Application
System
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 26
Case example
Metamodel and defined complexity aspects
Line of Business Business
Function Organization Unit
Application
Domain
Application
Component
Infrastructure
System
Technical
Domain
Ap
pli
cati
on
Arc
hit
ectu
re
Bu
sin
es
s
Arc
hit
ectu
re
Te
ch
nic
al
Arc
hit
ectu
re
Information Entity
3 4 5
6
7
8
Application
System 1
Application
Interface 2
Platform
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 27
Defined complexity aspects
Interpretation of number and heterogeneity (1)
Complexity Aspect Interpretation N Interpretation H
Application System
by Application Vendor
(Type 1)
Concentration of applications
by vendors
Decreases with a shift from special
to standard / mainstream vendors
Number of all existing
(productive) applications
Decreases with the decommissioning
of applications
Function
Implementation
by Application Name
(Type 2)
Concentration of function implemen-
tations by applications
Decreases with a shift of functions
from peripheral to core applications
Number of function implementations
through applications
Decreases with the removal of
function implementations
Application Usage
by Application Name
(Type 2)
Concentration of application usage
by applications
Decreases with a usage shift
from ‘local‘ to ‘global‘ applications
Number of application uses by
organization units
Decreases with the reduction of
application uses by organization units
Application Interface
by Interface Type
(Type 1)
Concentration of interfaces
by interface technologies
Decreases with a shift from special
to standard technologies
Number of all existing
application interfaces
Decreases with the decommissioning
of application interfaces
1
2
3
4
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 28
Defined complexity aspects
Interpretation of number and heterogeneity (2)
Complexity Aspect Interpretation N Interpretation H
LoB Support
by Application Name
(Type 2)
Concentration of LoB support
by applications
Decreases with a shift from LoB
specific to ‘global’ applications
Number of business line supports
by applications
Decreases with a reduction of busi-
ness line support by applications
Interface Usage
by Interface Name
(Type 2)
Concentration of interface usage
by interfaces
Decreases with a shift from special
to (reusable) standard interfaces
Number of interface usages
by applications
Decreases with a reduction of the
interface usage by applications
Platform Usage
by Platform Name
(Type 3)
Concentration of platform usage
by platforms
Decreases with a shift from special
to standard platforms
Number of platform usages by
applications
Decreases with a reduction of the
platform usage by applications
Information Usage
by Application Name
(Type 2)
Concentration of information usage
by applications
Decreases with a shift of data from
peripheral to core applications
Number of information object
usages through applications
Decreases with a reduction of inf.
object usage through applications
5
6
7
8
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 29
Measurement results
Development of complexity measures (from as-is to target)
Line of Business Business
Function Organization Unit
Platform
Information Entity
Application
Interface
Ap
pli
cati
on
Arc
hit
ectu
re
Bu
sin
es
s
Arc
hit
ectu
re
Te
ch
nic
al
Arc
hit
ectu
re
Application
System
7%
-18%
-32% -20% -27% -19%
-54%
-20%
-55%
-21%
Legend
N (delta in %)
H (delta in %)
-3%
-80%
68%
-7%
-34% -27%
5 3 4
1
8
6
7
2
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 30
Results
Interpretation of measure variation (1)
Complexity Aspect As-is / to-be Interpretation
Application System
by Application Vendor
(Type 1)
Elimination of EUC components and special
solutions results in a substantial reduction of N
H remains high as the target architecture is still
mainly based on individual applications
(51; 3.90) →
(23; 3.08)
Function
Implementation
by Application Name
(Type 2)
Functional disentanglement (mainly in Technical
Accounting) results in considerable reduction of N
Centralization of functionality in few core applications
results in a reduction of H
(148; 3.09) →
(108; 2.50)
Application Usage
by Application Name
(Type 2)
Application consolidation (mainly EUC) results in a
substantial reduction of N
Elimination of local applications (e. g., GL) results in
minor reduction of H (regional consolidation)
(65; 3.88) →
(44; 3.12)
Application Interface
by Interface Type
(Type 1)
Interface consolidation and introduction of new
interfaces keep the balance
Introduction of centralized integration platforms
(EAI / ETL) results in a significant reduction of H
(40; 3.36) →
(39; 0.66)
1
2
3
4
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 31
Results
Interpretation of measure variation (2)
Complexity Aspect As-is / to-be Interpretation
LoB Support
by Application Name
(Type 2)
Application consolidation (mainly EUC) results in a
substantial reduction of N
Elimination of LoB specific applications (e. g., in
Exposure Management) results in reduction of H
(91; 3.90) →
(42; 3.11)
Interface Usage
by Interface Name
(Type 2)
Introduction of new applications and interfaces
(automation) results in an increase of N
Increased reuse of application interfaces results in a
minor reduction of H
(40; 3.69) →
(67; 3.45)
Platform Usage *
by Platform Name
(Type 3)
Introduction of new applications (e. g., CRM,
Solvency II) results in minor increase of N
Technical standardization of applications results in
significant reduction of H
(29; 2.18) →
(31; 1.78)
Information Usage
by Application Name
(Type 2)
Application consolidation and disentanglement of
data usage results in considerable reduction of N
Concentration of data usage in core applications incl.
reporting warehouse results in reduction of H
(182; 3.33) →
(120; 2.43)
5
6
7
8
*) excluding client platforms
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 32
Agenda
1. The Role of Complexity
2. Measuring EA Complexity
3. Case Example: Application Landscape Planning
4. Lessons Learned and Recommendations
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 33
Lessons learned
Evaluation by project participants
The calculated measures matched very well with the qualitative judgments of the architects involved.
In summary, the approach was rated as follows:
holistic / multidimensional approach
complete coverage of all metamodel aspects
flexible application at all architecture layers /
levels of detail incl. possibility to drill-down
unified interpretation and (potentially) simplified
aggregation of measures
complete metamodel- und tool-neutrality
simple application based on existing
architecture data
higher precision through the use of
concentration measures
focus on structural complexity only
modularity (one of the major concepts to
master complexity) is difficult to capture
complexity aspects need to be defined and
interpreted carefully
there is a certain risk that complexity is
optimized ‘out of sight’ (e. g., by moving it to
the subsystem level; this can be handled by
extending the model view appropriately)
global minimization of complexity may lead to a
local increase (which can be detected by
means of a drill-down and must be handled
accordingly)
Benefits: Drawbacks / to be noted:
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 34
Possible extensions
Adding measures from network analysis
Centrality
Closeness Centrality: shortest path of a
node to all other nodes
Betweeness Centrality: number of shortest
paths crossing a given node
Eigenvector Centrality: extent that a node
is connected to other central nodes
Modularity
Modularity: number of edges that fall into
groups minus the expected number in an
equivalent network with random edges
Cluster Coefficient: average share of
neighbours of a node, which are neighbours
themselves
Measures from the domain of Network Analysis may be a good complement to the CEAR approach
[SF12]. They are well suited to analyze the communication between active system elements
(e. g., applications and interfaces).
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 35
Application scenarios
Possible applications for complexity measures
1. Decision support / simulation for (major) architecture decisions (e. g., target application
landscape planning and landscape transformation)
2. Comparative analyses within an organization (e. g., to explain differences in costs and
flexibility)
3. Cross-organizational architecture benchmarkings (complementing existing cost
benchmarkings, etc.)
4. Systematic planning of target values as part of a continuous architecture management
(e. g., differentiated by architecture layers and domains)
5. Data delivery to IT risk management and the management of operational risks
(both, complexity models and measures)
…
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 36
General recommendations
Complexity measurement and management
1. Stakeholder Orientation
the stakeholders of complexity analyses and
their respective concerns should be
identified from the very beginning
the calculation of measures needs to be
aligned with stakeholder requirements
complexity assessments should be
performed on the back of management
questions / drivers only
comparative as-is analyses (e. g., between
different business domains) should not be
conducted without a clear top management
mandate
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 37
General recommendations
Complexity measurement and management
2. Rigorous Data Management
the reporting should be based on the EA
repository only and be fully automated
all stakeholders need to be informed about
the impact of architecture data on the
measures and be in a position to update that
data in a timely manner
a method needs to be defined of how to deal
with missing data in a reliable and uniform
way
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 38
General recommendations
Complexity measurement and management
3. Qualitative Analysis and Interpretation
complexity assessments require in-depth
analyses, which can only by carried out by
skilled and experienced architects
the analysis often requires drill-down
operations and supplemental research
the results of complexity assessments should
always be commented and interpreted
qualitatively (e. g., in a report)
architecture decisions should never be
based on complexity measures alone
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 39
General recommendations
Complexity measurement and management
4. Compensation for Local Complexity Increase
the global minimization of complexity may
lead to an increase on the local level
in particular, application consolidation will
usually result in increasing dependencies
of the target system thereby causing higher
coordination and change efforts
this should be detected in advance using
appropriate analysis methods (e. g.,
drill-down)
for the implementation to be successful,
the resulting changes need to be
communicated properly and additional
resources be provided
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 40
General recommendations
Complexity measurement and management
5. From Business to IT Architecture
to achieve maximum effectiveness,
complexity optimizations should begin with
the business
this will not be possible without a respective
problem awareness in the business and the
willingness to act (esp. in the top
management)
a comprehensive approach to business
architecture management is the ideal
means to achieve this
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 41
Outlook
Still a lot of things to figure out
…but the time seemed
right to make a start
A long way of
research ahead…
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 42
References
[IEE11] IEEE Computer Society (Eds.): Systems and Software Engineering — Architecture Description
(ISO/IEC/IEEE 42010:2011). IEEE Computer Society, New York, 2011.
[JB79] Jacquemin, A. P.; Berry, C. H.: Entropy Measure of Diversification and Corporate Growth. The Journal of
Industrial Economics (27:4), 1979; S. 359-369.
[Le97] Lehman, M. M.: Laws of software evolution revisited. In: (Montangero, C. Eds.): Software Process
Technology – Proceedings of the 5th European Workshop on Software Process Technology (EWSPT ‘96),
Nancy, 1996. Springer, Berlin; S. 108–124.
[Ro79] Ropohl, G.: Eine Systemtheorie der Technik – Zur Grundlegung der Allgemeinen Technologie. Carl Hanser,
München, 1979.
[Ru13] Rutz, U.: IT Complexity Management @ Commerzbank – Overview on Commerzbank Initiative.
Präsentation EAMKON Konferenz, Stuttgart, April 2012.
[SB11] Schmidt, C.; Buxmann, P.: Outcomes and Success Factors of Enterprise IT Architecture Management:
Empirical Insight from the International Financial Services Industry. European Journal of Information
Systems (20), 2011; S. 168-185.
[Sc13] Schneider, A.: CALM3 – Workshop 3. Präsentation beim 3. Workshop CALM3 (Complexity of Application
Landscapes – Models, Measures, Management). Technische Universtität München, August 2013.
[Sh48] Shannon, C. E.: A Mathematical Theory of Communication. Bell System Technical Journal, 27, 1948; 379-
423 und 623-656.
[SM03] Schneberger, S. L.; McLean, E. R.: The Complexity Cross: Implications for Practice. Communications of the
ACM, 46 (9), 2003; S. 216-225.
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 43
References
[SF12] Simon, D.; Fischbach, K.: IT Landscape Management Using Network Analysis. In: (Poels, G. Ed.):
Enterprise Information Systems of the Future, 6th IFIP WG 8.9 Working Conference, CONFENIS 2012,
Springer, 2012; S. 18-34.
[St73] Stachowiak, H.: Allgemeine Modelltheorie. Springer, Wien, New York, 1973.
[SWS13] Schmidt, C.; Widjaja, T.; Schütz, A.: Messung der Komplexität von IT-Landschaften auf der Basis von
Architektur-Metamodellen: Ein generischer Ansatz und dessen Anwendung im Rahmen der Architektur-
Transformation. Informatik 2013 – Beiträge der 43. Jahrestagung der Gesellschaft für Informatik. Köllen
Verlag, Bonn.
[SWK13] Schütz, A.; Widjaja, T.; Kaiser, J.: Complexity in Enterprise Architectures – Conceptualization and
Introduction of a Measure from a System Theoretic Perspective. European Conference on Information
Systems (ECIS), Utrecht, 2013.
[TOG13] The Open Group (Ed.): TOGAF (The Open Group Architecture Framework) V. 9.1.
http://www.opengroup.org/togaf, 2013.
[Wi12] Widjaja, T.; Kaiser, J.; Tepel, D.; Buxmann, P.: Heterogeneity in IT Landscapes and Monopoly Power of
Firms: A Model to Quantify Heterogeneity. International Conference on Information Systems (ICIS),
Orlando, 2012.
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 44
Developing
your core
competencies
About us
Our business segments
setup of EA practices
EA health checks /
realignment of EA practices
development of the business
architecture
introduction of EA tools
architect trainings
EA team coaching
capturing of as-is
landscapes
target landscape planning
management of
transformation projects /
programs
development of technical
standards and reference
architectures
development of solution
architectures
2. Enterprise Architecture
Office
implementation of
governance frameworks
(esp. COBIT)
development of the IT
strategy
setup of portfolio
management practices
optimization of the IT
organization
implementation of risk
management frameworks
3. IT-Strategie und
IT-Governance
Expanding
your operating
range
Aligning your
IT with the
business
2. Enterprise Architecture
Office
1. Development of EA
Capabilities
3. IT Strategy and IT
Governance
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 45
About us
Our publications
Dr. Christian Schmidt | How to Measure Enterprise Architecture Complexity | Open Group Conference | October 21st, 2013 46
About us
Memberships and research cooperation
Organizations:
Research Partners: Research Projects:
CEAR
Complexity in
Enterprise
Architectures
CALM³
Complexity of Application
Landscapes – Models,
Measures, Management
Communities:
EA Community Rhein-Main
Thank you for your attention