Goal Based Alignment of Enterprise Architectures
Balbir S. Barn and Tony Clark
School of Engineering and Information Sciences, Middlesex University, London NW4 4BT, U.K.
Keywords:
Enterprise Architecture, Simulation, Modeling.
Abstract:
Business and IT alignment remains an ongoing concern for organizations. In this paper, we propose a set of
technologies and concepts - notably goals and computable functions which can be used to provide a measure
of equivalence between as-is and to-be enterprise architectures.
1 INTRODUCTION
Business and IT alignment has remained an ongoing
concern for organisations since the 1980s (Luftman,
2004). Throughout this period, researchers have ad-
dressed the importance of alignment and in particu-
lar the need for congruence between business strat-
egy and and IT strategy (Chan and Reich, 2007). The
Strategic Alignment Model (SAM) (Henderson and
Venkatraman, 1993). defines that alignment is the de-
gree of fit and integration among business strategy, IT
strategy, business infrastructure, and IT infrastructure.
Enterprise Architecture (EA) aims to capture the
what, why and how of a business including the align-
ment between business functions and IT systems. De-
spite the opportunities presented for addressing Busi-
ness and IT Alignment (BIA) using EA, the current
state of the art of EA presents issues such as: large
unwieldy methods such as those derived from TO-
GAF (Spencer et al., 2004); a lack of precision in the
methods. This paper contributes a rigorous approach
to aligning business goals to IT infrastructure using
functions that can be applied to both as-is and to-be
enterprise architecture models.
2 RELATED WORK
A model for measuring BIA was originally proposed
by Henderson and Venkatraman, the Strategic Align-
ment Model (SAM) (Henderson and Venkatraman,
1993). Other researchers have extended the SAM,
e.g. Luftman proposed the Strategic Alignment Ma-
turity Model (SAMM) measures maturity levels of
strategic alignment along several key dimensions such
as governance and skills. and categorizes levels
into strategic, tactical and operational states (Luft-
man, 2004). Vargas et al. have produced a model
that has been consolidated from a review of exist-
ing models and have studied the use of the model in
case studies of public universities in Nicaragua (Var-
gas Chevez, 2010). Other models for strategic align-
ment have focused on executive feedback measures
(Avison et al., 2004), organizational structure (Berg-
eron et al., 2004), social dimensions such as vision
(Reich and Benbasat, 1996) and cognitive dimensions
(Tan and Gallupe, 2006). Chung et al. recommend
that new research should examine the recursive rela-
tionship between alignment and the extent of appli-
cations implementation and IT infrastructure flexibil-
ity (Chung et al., 2003). Street has begun this work,
examining ‘service gaps’ as a measure of alignment
(Street and Denford, 2012). One of the emerging stan-
dardisation efforts on goal modelling is the Business
Motivation Model (BMM)
1
. This model provides a
set of concepts for describing business plans includ-
ing links to other notations such as BPMN.
3 LEAP
The LEAP modelling approach is an integrated
model, method and simulation environment that uses
concepts from component based design, event driven
architecture and service oriented architecture. The ap-
proach has been described in detail elsewhere (Clark
et al., 2011; Clark and Barn, 2011; Clark and Barn,
2012).
In order for an organization to understand the im-
pact of a business goal both in terms of current tech-
1
http://www.omg.org/spec/BMM/1.1/PDF/
230
S. Barn B. and Clark T..
Goal Based Alignment of Enterprise Architectures.
DOI: 10.5220/0004081802300236
In Proceedings of the 7th International Conference on Software Paradigm Trends (ICSOFT-2012), pages 230-236
ISBN: 978-989-8565-19-8
Copyright
c
2012 SCITEPRESS (Science and Technology Publications, Lda.)
nology and likely changes, the organization must first
describe the business goal using the LEAP DSL. The
goal is described using three basic elements: (1) a de-
scription of the goal and quality attribute that captures
the essence of the goal; (2) a set of assumptions that
are either invariant or are values that are indicative of
the goal being met; (3) a function F from system exe-
cution traces to a partially ordered domain that can be
used as a measure of goal satisfaction.
Our proposal is that architectural alignment with
respect to a business goal must be expressed in terms
that can be precisely measured. Our claim is that
any meaningful business goal can be recast as a func-
tion over the architectures; if the values produced by
the function are ordered such that the ordering corre-
sponds to goal satisfaction then we can use this to test
whether the to-be architecture is in some way better
than the as-is architecture with respect to the business
goal.
4 CASE STUDY AND
EVALUATION
Higher education institutions (HEI) in the UK are
faced with a challenging and dynamic business en-
vironment where public funding of HEIs has been
reduced by up to 70%. This lost funding is being
replaced by the introduction of a new student fees
regime beginning in 2012 following a bill introduced
in the UK parliament in November 2010. This en-
vironment places greater expectations on operational
efficiency and one mechanism for addressing this is
to ensure that business strategy and IS/IT strategy is
better aligned (Gregor et al., 2007).
The University of Scrabbleshire (UoS), has de-
cided that they will introduce student fees at the maxi-
mum permitted level of £9000. However, they recog-
nise that such high fees will have to be justified so
the Corporate Plan has introduced new requirements.
The University will compete on quality in terms of re-
search outputs. Such a change will hopefully lead to
a higher ranking in the research league tables.
To address these business goals, UoS recognises
that it needs to go through a number of stages, firstly it
will have to assess the current systems and processes
and how they support the business goals. Some mea-
sure of “quality” or “extent” of how these goals have
been met will have to be derived. Currently, there is
no mechanism for approaching this measurement ca-
pability. A second stage is to identify changes to the
systems, perhaps identify new systems and or busi-
ness processes. A third stage is of course to measure
how how the proposed changes have addressed these
business goals.
During the assessment of the existing systems, the
IT services department working under the leadership
of Dr Magnus Sungam, identify the following sys-
tems that are potentially impacted in some form by
the new corporate plan: (1) timetabling systems that
schedule staff to specific modules to specific room (2)
the workload management system used by Heads of
Department to ensure that staff are allocated work in
an appropriate manner; (3) the University Repository
that provides external access to the research outputs
of the University; (4) the HR system that manages
contracts for staff; (5) the University Website.
The various systems identified need to change in a
number of ways in order to support the business goals
laid out in the corporate plan. Various changes are en-
visaged. The workload management system will need
to differentiate and support workload planning for dif-
ferent categories of staff (those that are research ac-
tive and those that are teaching only). The timetabling
system will need to integrate new constraints that en-
sure that research active staff have at least and prefer-
ably three clear days to support research activity. Both
these systems will need to extract information from
the HR contracts system. The University Repository
will need to be integrated with the external facing
Corporate website so that changes to an academic’s
research outputs is updated automatically on the web-
site to reflect updates to the Repository. A new system
to manage the research process for preparing to the
UK national research assessment process for ranking
UK universities will need to be developed.
5 ASSUMPTIONS
There are a collection of business assumptions and di-
rectives that must be defined before the goal can be
represented and analysed. A directive is something
that must hold at all times, for example: (1) In plan-
ning any changes to UoS no new staff resources can
be made available; (2) All changes must maintain the
current levels of teaching.
Each assumption allows us to implement a change
without taking the wider context into account when
analyzing or measuring the effect of the change in
terms of the overall goals, for example: (1) Reputa-
tion will be enhanced by visibility of research outputs
via the UoS web-site; (2) UoS is effective at measur-
ing the quality of its research outputs; (3) Research
is to be measured on a 4-point scale with 1 being the
lowest level of quality where the likelihood of pro-
ducing high-quality research is increased by provid-
ing staff with a contiguous period of time free from
GoalBasedAlignmentofEnterpriseArchitectures
231
teaching and admin responsibilities; (4) The number
of research outputs and their quality are positively
correlated to reputation.
6 SPECIFICATION: AS IS
The LEAP component specifications provide opera-
tions that allow the UoS to be simulated in terms of its
research activities. This section provides an overview
of the UoS LEAP specification as a single component
scabbleshire that contains sub-components and an
operation perform that is used to define the simula-
tion:
component s cr a b bl e s h ir e {
component p er s on n e l { model { class S ta ff { n am e :str } } }
The first component, shown above, is personnel that
maintains a database of all staff employed by the uni-
versity. Currently UoS employs all academic staff on
the same type of contract. The next component is used
to record research activity:
component r e se a r ch _ a c ti v i t y {
model { class A c ti v it y { s ta f f :str; q u al i ty :int; j o ur n al :str}}
port a c ti o ns [in]: interface {
st a r t _r e s ea r c h ( st af f :str, jo u rn a l :str):void;
pe r f o rm _ r e s e a r ch ( s ta f f :str):void;
ab o r t _r e s ea r c h ( st af f :str):void }
port p r od u ce [ o u t ]: interface {
re s ea r c h ( st a f f :str, qu al i ty :int, j ou r na l :str):void }
spec {
ab o r t _r e s ea r c h ( na me :str):void {
pre S ta f f ( n ame , qua li ty , j o ur n al ) ? ( q ua li ty > 0)
post S t af f ( n ame ,0 )
messages p r od u ce < - r e se a rc h ( nam e , q u al it y , jo u rn a l ) }
st a r t _r e s ea r c h ( na me :str, jo u rn a l :str):void {
pre not( St af f ( n ame , _ , _ ) )
post S t af f ( n ame ,0 , j o ur n al ) }
pe r f o rm _ r e s e a r ch ( n am e :str):void {
pre S ta f f ( n ame , a1 , j o ur n al ) ? ( a1 < 3)
post S t af f ( n ame , a2 , j o ur n al ) ?( a 2 = a1 +1) }
pe r f o rm _ r e s e a r ch ( n am e :str):void {
pre not( St af f ( n ame , a1 , j o ur n al ))
post S t af f ( n ame ,0 , j o ur n al ) }
pe r f o rm _ r e s e a r ch ( n am e :str):void {
pre S ta f f ( n ame , 3 , jo ur n al )
post not( S ta f f ( n ame , _ , _ ))
messages p r od u ce < - r e se a rc h ( nam e ,4 , jo u rn a l ) } } }
The research activity component manages a
database that maintains information about staff re-
search activity. An Activity record includes the
quality of the research (on a scale of 0 to 4) and the
Journal to which it will be submitted if completed.
The input interface actions support operations for
starting a research activity, performing research and
aborting the research. After initiating a piece of work,
a member of staff can perform up to 4 increments at
which point the research is ready for submission. At
any time the research can be aborted, at which time it
is assumed to be submitted as-is providing some work
has been undertaken.
A head of department manages a resource
database that allocates each member of staff to teach-
ing, research and administration tasks:
component r e so u r ce _ p l an n i n g {
model {
class St a ff { na me :str; t ea c h in g :int; re s ea r ch :int;
ad mi n :int} }
invariants {
ti m e_ 1 0 0 {
forall S ta ff (_ , te ac hi ng , r es ea rc h , ad m i n )
in state { t ea c hi n g + re s e a r c h + a d mi n = 1 00 } } } }
The UoS uses a centralized room booking system.
The simulation uses the room booking system to de-
termine whether a member of staff can undertake re-
search:
component r oo m _ bo o k in g {
model { class B o ok i ng { s ta f f :str; r oo m :int; t im e :int} } }
The UoS research repository contains a record of re-
search outputs produced by members of staff. We will
assume that all submissions added to the repository
are in print and that the research quality depends on
the length of time spent on the research. The opera-
tion research is used to update the repository:
component r ep o s i t o ry {
model {
class En t ry { st a ff :str; en tr y :int; qu a li t y :int; j o ur n al :str}
}
port d e po s it [in]: interface {
re s ea r c h ( n a me :str, qu a li t y :int, jo u rn a l :str):void }
spec {
re s ea r c h ( n a me :str, qu a li t y :int, jo u rn a l :str):void {
post
En tr y (n ame , ent ry , qua li ty , j o ur n al )
not( En tr y ( _ , e ntr y , _ ) in state@pre) } } }
The UoS web site is the public facing interface for
UoS. It contains many types of entry but for the
purposes of the simulation it is important to know
whether a repository entry is referenced on the web
site:
component web { model { class En tr y { r ep o s it o r y_ i d :int} } }
The scrabbleshire component is completed by
defining the data and operations used for the simu-
lation. The simulation is driven by a collection of
time ordered messages that are delivered to the sub-
component defined above. Each message has a time,
name and some argument data:
model {
class M es s ag e { ti me :int; n am e :str; a rg s : [ V a lu e ] }
class Ti me { t im e :int } }
The simulation is executed by the perform operation
that delivers messages. A message is ready for de-
livery when its time becomes current, if there are no
messages ready then time advances:
spec {
pe r fo r m ():void {
pre T im e ( t ) not( Me s sa g e ( t , m es sa ge , a rg s ))
post T im e ( t t ) ?( tt = t + 1 ) } }
.. . f u rt h er sp e ci f i ca t i on s b e lo w ...
There are three different messages: start that at-
tempts to initiate research by a member of staff;
do research that attempts to make an incremental
research step; transfer that ensures that all reposi-
tory entries for a member of staff are transferred to the
ICSOFT2012-7thInternationalConferenceonSoftwareParadigmTrends
232
UoS web site. When research starts the appropriate
message is sent to the research activity module:
pe r fo r m ():void {
pre T im e ( t ) M e ss a ge (t , ’ st ar t ’ ,[ na me , jo ur n al ])
post not( M e ss a ge (t , ’s t ar t ’ ,[ na me , jo ur n al ] ))
messages r e s e a rc h _ a ct i v i ty . a c ti o ns
<- st a r t _ r e s ea r c h ( n ame , jo u rn a l ) }
When the simulation attempts to direct a member of
staff to do research it may be prevented from succeed-
ing because they are timetabled for teaching, other-
wise a message is sent to the research activity
component:
pe r fo r m () :void {
pre T im e ( t ) M e ss a ge (t , ’ do _ r es e a rc h ’ ,[ nam e ])
?( !exists St af f ( name , _ , t ) in r o om _ b oo k i ng .state)
post not( M e ss a ge (t , mes sa ge , a rg s ))
messages r e s e a rc h _ a ct i v i ty . a c ti o ns < - p e r fo r m _ re s e ar c h ( na
me )
}
pe r fo r m ():void {
pre T im e ( t ) M e ss a ge (t , ’ do _ r es e a rc h ’ ,[ nam e ])
?(exists St a ff ( nam e ,_ , t ) in r o om _ b oo k i ng .state)
post not( M e ss a ge (t , mes sa ge , a rg s )) }
When the simulation attempts to direct a member of
staff to transfer their research from the repository to
the UoS web site, teaching can prevent the transfer:
pe r fo r m ():void {
pre T im e ( t ) M e ss a ge (t , ’ tr a ns f er ’ ,[ nam e ])
?( !exists St af f ( name , _ , t ) in r o om _ b oo k i ng .state)
post ?(forall E nt r y ( n ame , id , _) in r e po s i to r y .state {
exists E nt ry ( id ) in we b .state }) }
pe r fo r m ():void {
pre T im e ( t ) M e ss a ge (t , ’ tr a ns f er ’ ,[ nam e ])
?(exists St a ff ( nam e ,_ , t ) in r o om _ b oo k i ng .state)
post not( M e ss a ge (t , mes sa ge , a rg s )) }
Our assumption is that effective research needs a con-
tiguous amount of free time slots. Therefore, when
staff are scheduled for teaching, current research is
aborted (i.e. submitted as-is):
pe r fo r m ():void {
pre T im e ( t ) S ta f f ( n ame , _ , t ) in r o om _ b oo k i ng .state
messages ro o m _b o o ki n g . a ct i on s <- a bo r t _ r e s e ar c h ( na me ) }
The global invariants ensure that all staff managed by
the resource planning and room booking components
are registered with personnel:
invariants {
re s o u rc e d _s t a f f {
forall S ta ff ( nam e ,_ ,_ , _ , _ )
in r e so u r c e_ p l an n i n g .state {
exists S ta ff ( n am e ) in p e rs o n n e l .state } }
bo o k ed _ s ta f f {
forall S ta ff ( nam e ,_ ,_ ) in r oo m _ bo o k in g .state {
exists S ta ff ( n am e ) in p e rs o n n e l .state } } }
Finally, the components are connected in the initial-
ization clause of scrabbleshire:
init { c o nn e ct ( r e s ea r c h _a c t i vi t y . pr od uc e , re p os i t or y . d ep o si t )}
7 SEMANTICS
The operational semantics of a LEAP specification is
defined in figure 1. A component κ is represented as
(c,∆,S) where c is the component identifier, ∆ is the
component database and S is a set of operation speci-
fications. LEAP allows components to be nested; the
specification flattens this tree structure. LEAP sup-
ports invariants on each component that are assumed
to hold at all times within a component κ. Finally,
LEAP ports are not represented in the semantics and
port connections are handled by directing output mes-
sages to the target component.
An operation specification m(
˜
i)
µ
p,q
∈ S consists of
an operation name m, a sequence of formal parame-
ter names
˜
i, a precondition p, postcondition q and a
predicate µ. The specification requires that when the
operation named m is invoked in response to process-
ing a message with the same name, if p is true of the
pre-state of the component then q is true of the post-
state and µ defines the messages that are sent.
The semantics in figure 1 defines a relationship
M ` C ⇒ C
0
,M
0
that defines an execution of com-
ponents C with input messages M producing a new
collection of components C
0
with a mixture of unpro-
cessed messages and new output messages M
0
. Rule
L-1 is the workhorse that defines how a single step
is performed by processing a message. The precon-
dition must hold for the pre-state ∆ in the context of
the supplied argument values ˜v and values for the free
variables FV(p), the postcondition must hold for both
the pre-state (referenced as state@pre) and the post-
state ∆
0
with respect to the argument values, the same
values used for free variables in the precondition, and
values for the free variables in q. Finally the message
predicate µ must hold for the output messages in the
context of all the previous variable bindings.
Rule L-2 just lifts the single step defined in L-1 to
sequences of messages and sets of components. Note
that the new output messages in L-2 (M
1
) are added to
the end of any unprocessed messages (M
0
). Rule L-2
defines how sequences of execution are concatenated.
We define a relationship C
0
→
M
T that holds be-
tween a collection of UoS components C
0
, a set
of message terms M = {Message(t, n,a), ...} and
an execution trace T = [C
0
,C
1
,. ..,C
n
] such that
[perform()] ` C
0
⇒ C
n
,
/
0 and all messages M have
L-1
p[ ˜v
1
/
˜
i][ ˜v
2
/FV (p)](∆)
q[ ˜v
1
/
˜
i][ ˜v
2
/FV (p)][ ˜v
3
/FV (q)](∆, ∆
0
)
µ[ ˜v
1
/
˜
i][ ˜v
2
/FV (p)][ ˜v
3
/FV (q)][ ˜v
4
/FV (µ)](∆, ∆
0
,M)
m
c
( ˜v
1
) ` (c,∆, m
µ
(p,q)
(
˜
i) : S) ⇒ (c,∆
0
,m
µ
(p,q)
(
˜
i) : S),M
L-2
m ` κ ⇒ κ
0
,M
1
m : M
0
` {κ} ⇒ {κ
0
},M
0
+ M
1
L-3
M
1
` C
1
⇒ C
2
,M
2
M
2
` C
2
⇒ C
3
,M
3
M
1
` C
1
⇒ C
3
,M
3
Figure 1: LEAP Semantics.
GoalBasedAlignmentofEnterpriseArchitectures
233
Figure 2: Goal function.
been processed.
8 GOAL FUNCTION
The UoS business goal can be implemented as a mea-
sure F on execution traces: (F
1
) the number of re-
search outputs in the UoS repository, our assumption
is that it is necessary to increase the number of outputs
in order to increase reputation; (F
2
) the weighted total
number of research outputs, reputation is increased
by maximizing the locally attributed quality of each
output; (F
3
) maximizing staff engaged in research de-
composing into F
3a
and F
3b
; (F
3a
) the total number
of academic staff engaged in research; (F
3b
) the num-
ber of times that each member of academic staff has
an opportunity to produce an output of the maximum
level of quality; (F
4
) minimizing the time spent to pro-
duce a given number of outputs.
The measure F is defined in figure 2 where the
following assumptions are made: R=repository and
P=personnel; # produces the size of a set or se-
quence; sequences (of messages) are concatenated
using +; the set ∆
i
is the set of terms of the form
Entry(n,e,i, j); the set S used in the definition of F
3b
is the set of all staff names. Given a simulation de-
fined by a collection of messages M and a UoS speci-
fication C then C →
M
T . The measure of how well C
meets the business goal is represented by F(T ).
C
a
→
M
T
a
-
C
b
→
M
T
b
γ
a
F
?
≤
-
γ
b
F
?
Figure 3: Goal alignment.
9 CONFORMANCE
Figure 3 shows how this machinery is intended to sup-
port the comparison of business architectures, by pro-
ducing a measure γ
a
for a specification C
a
(the as-is
architecture) and a measure γ
2
for the specification C
b
(the to-be architecture), for the same simulation mes-
sages M. Since each measure is an integer, they can
be compared with ≤ and we claim that C
b
is better
at satisfying the business goal than C
a
is the diagram
commutes.
10 SPECIFICATION: TO BE
The model for personnel is updated and staff are
moved on to different types of contracts. Staff are ei-
ther research active, in which case they are expected
to spend at least 50% of their time on research ac-
tivities, or are teaching only in which case they are
allocated no time for research:
model {
class St a ff { na me :str }
class C on t ra c t { i d :int }
class T ea c h in g O nl y extends C on t ra c t { }
class R e s e a r ch A c ti v e extends Co n tr a c t { }
assoc Le g al { s ta ff S t af f c on t ra c t C on t ra c t } }
The repository is updated to generate an event each
time an update is made. Any UoS component can
monitor the event stream in order to detect the update:
component r ep o s i t o ry {
model {
class En t ry { st a ff :str; en tr y :int; qu a li t y :int; j o ur n al :str}
}
port d e po s it [in]: interface {
re s ea r c h ( n a me :str, qu a li t y :int, jo u rn a l :str):void }
port e v en t s [ o ut ]: interface {
up d at e ( en tr y :int):void }
spec {
re s ea r c h ( n a me :str, qu a li t y :int, jo u rn a l :str):void {
post E n tr y ( n ame , ent ry , qua li ty , j o ur n al )
not( En tr y ( _ , e ntr y , _ ) in state@pre)
messages e v en t s < - u p da te ( e nt r y ) } } }
The web site is updated to monitor respository events:
component web {
model { class E nt r y { r e p os i t o ry _ i d :int } }
port m o ni t or [in]: interface { up d at e ( en tr y :int):void }
spec { u pd a te ( e nt ry :int):void { post E n tr y ( en tr y ) } } }
A new component is introduced that manages a
database of quality measures for journals. This will
be used to override local quality measures for research
outputs:
component q u a li t y {
model { class J o ur n al { n am e :str; q u al i ty :int } } }
A global invariant is added that enforces the UoS di-
rective that teaching only staff wil not be given time
for research activities:
te a c hi n g _ on l y {
forall L eg al ( S ta ff ( n am e ) , Te a ch i n gO n l y ( _ ) )
in p er s on n e l .state {
exists S ta ff ( nam e ,_ ,0 , _ ) in r e s ou r c e_ p l a nn i n g .state
}
}
Similarly, research active staff must be given at least
50% of their time for research:
ICSOFT2012-7thInternationalConferenceonSoftwareParadigmTrends
234
re s e a rc h _ ac t i v e {
forall L eg al ( S ta ff ( n am e ) , Re s e ar c h Ac t i ve ( _ ))
in p er s o n n e l .state {
exists S ta ff ( nam e ,_ , re se ar c h , _ )
in r e so u r c e_ p l an n i n g .state { r e se a r c h >= 50 } } }
Finally, all research active staff should be given 3 con-
tiguous days free of teaching where this is possible:
da y s _ fo r _ r es e a r c h {
forall L eg al ( S ta ff ( n am e ) , Re s e ar c h Ac t i ve ( _ ))
in p er s o n n e l .state {
forall B oo k in g ( name ,_ , t )
in r oo m _ b o o k in g .state {
let w ee k = [ t im e |
Bo o ki n g ( n ame , _ , tim e ) <- r o o m _ bo o k in g s .state,
?( ti me >= t a nd t im e <= t +5) ]
in if ( le n gt h ( we ek )) = 2
then
case bo o k in g s _ fo r _ we e k {
[x , y ] -> a bs (x - y ) = 1 or ab s ( x - y ) = 4
}
else false } } }
The quality measure (∆
i
in F
2
) is now redefined so
that it is the maximum of the locally defined quality
and the independently defined quality of the journal.
11 VERIFICATION
Section 8 defines the function and this section
uses rigorous argument to establish the following
proposition: Given the UoS architectures, C
a
and C
b
described in sections 6 and 10 respectively, then for
any collection of simulation messages M for which
C
a
→
M
T
a
and C
b
→
M
T
b
, and under the assumption
that both architectures use the same resources in
which staff who are not research active can be
identified, then F(T
a
) ≤ F(T
b
).
Proof: by case analysis on the components of
F: (F
1
) since C
b
includes an invariant constraint that
research active staff are given time to pursue research
where possible, then the number of research outputs
will be at least the same as in C
a
; (F
2
) in C
a
, ∆
i
is
defined in terms of a local measure for research out-
put quality and in C
b
, ∆
i
is defined as the maximum
of the local measure and the independent measure of
journal quality, therefore the weighted sum of output
quality will at least stay the same; (F
3a
) the number
of research active staff is the same as defined in the
proposition above; (F
3b
) the amount of time given
to all staff to undertake research is opportunistic in
C
a
and is managed in C
b
; under the assumption that
the manages process will identify contiguous time
for research wherever this is possible then the time
available to research active staff will be maintained
or increase in C
b
; (F
4
) the transfer from repository to
web in C
a
is opportunistic and is systematic in C
b
,
therefore the externally visible research profile of
UoS will be maintained or increased in C
b
compared
to C
a
. Therefore F(T
a
) ≤ F(T
b
) QED.
12 CONCLUSIONS
This paper has proposed an approach to EA alignment
with respect to business goals that involves the use of
precise architecture definitions that support goal mea-
surement functions. We have reviewed the literature
in this area and argued that existing technology for
EA is not sufficiently precise to support such an ap-
proach. We have used the LEAP modelling technol-
ogy to validate the approach with a real-world case
study and have shown that precisely defined architec-
ture models allow business goals to be represented as
computable functions and therefore architecture mod-
els to be compared with respect to goal satisfaction.
We aim to take this approach further by integrat-
ing LEAP with formal methods technologies such
as SAT solvers and model checkers in order to de-
termine whether EA alignment and goal satisfaction
modelling can be automated.
REFERENCES
Avison, D., Jones, J., Powell, P., and Wilson, D. (2004). Us-
ing and validating the Strategic Alignment Model. The
Journal of Strategic Information Systems, 13(3):223–
246.
Bergeron, F., Raymond, L., and Rivard, S. (2004). Ideal pat-
terns of strategic alignment and business performance.
Information & Management, 41(8):1003–1020.
Chan, Y. and Reich, B. (2007). IT alignment: what have
we learned? Journal of Information Technology,
22(4):297–315.
Chung, S., Rainer Jr, R., and Lewis, B. (2003). The impact
of information technology infrastructure flexibility on
strategic alignment and applications implementation.
Communications of the Association for Information
Systems, 11(11):191–206.
Clark, T. and Barn, B. (2011). Event driven architec-
ture modelling and simulation. In Service Oriented
System Engineering (SOSE), 2011 IEEE 6th Interna-
tional Symposium on, pages 43 –54.
Clark, T. and Barn, B. S. (2012). A common basis for mod-
elling service-oriented and event-driven architecture.
In Proceedings of the 5th India Software Engineering
Conference, ISEC ’12, pages 23–32, New York, NY,
USA. ACM.
Clark, T., Barn, B. S., and Oussena, S. (2011). Leap: a
precise lightweight framework for enterprise architec-
ture. In Proceedings of the 4th India Software En-
gineering Conference, ISEC ’11, pages 85–94, New
York, NY, USA. ACM.
Gregor, S., Hart, D., and Martin, N. (2007). Enterprise
architectures: enablers of business strategy and is/it
alignment in government. Information Technology &
People, 20(2):96–120.
GoalBasedAlignmentofEnterpriseArchitectures
235
Henderson, J. and Venkatraman, N. (1993). Strategic
alignment: Leveraging information technology for
transforming organizations. IBM Systems Journal,
32(1):4–16.
Luftman, J. (2004). Assessing business-IT alignment ma-
turity. Strategies for information technology gover-
nance, page 99.
Reich, B. and Benbasat, I. (1996). Measuring the linkage
between business and information technology objec-
tives. MIS Quarterly, pages 55–81.
Spencer, J. et al. (2004). TOGAF Enterprise Edition Version
8.1.
Street, C. and Denford, J. (2012). Punctuated equilibrium
theory in is research. Information Systems Theory,
pages 335–354.
Tan, F. and Gallupe, R. (2006). Aligning business and in-
formation systems thinking: A cognitive approach.
Engineering Management, IEEE Transactions on,
53(2):223–237.
Vargas Chevez, N. (2010). A unified strategic business and
IT alignment model: A study in the public universi-
ties of Nicaragua. In AMCIS. AIS Electronic Library,
KTH.
ICSOFT2012-7thInternationalConferenceonSoftwareParadigmTrends
236
