A Framework for A Business Intelligence-Enabled Adaptive ......BI-EAEA Approach •The Business...

A Framework for A Business Intelligence-Enabled Adaptive

Enterprise Architecture

Okhaide Akhigbe, Daniel Amyot and Gregory Richards

okhaide@uottawa.ca

Business IT Alignment

• Aligning business objectives with Information Systems has always presented a challenge.

• Is the blame on the Information Systems?

• Survey shows this is among the top priorities of CIOs today.

ER2014 p.2

ER2014 p.3

Source: IT WORLD CANADA (OCTOBER 17th 2014) http://www.itworldcanada.com/blog/the-alignment-of-it-and-business-doesnt-make-sense-so-dont-say-

it/98340

Source: InformationWeek (September 18th 2014) http://www.informationweek.com/strategic-cio/team-building-and-staffing/it-business-alignment-enough-already/a/d-id/1315851

Enterprise Architecture

• Informs how information, processes and technology should be positioned in an organization.

• Enterprise Architecture is supposed to offer alignment.

• Frameworks exist. Are they adaptive?

ER2014 p.4

ER2014 p.5

Current Business

Requirements

New Business

Requirements

Business evolves

Business

Objectives Information

System

Business requirements

drive changes Architecture drives

changes

Business

Objectives Information

System

Business Intelligence Theme

• BI is about getting data in and out. But to do what?

• BI helps organizations derive meaning from data (Useful in dynamic environments).

• BI Theme refers to instrumenting of data used to make decisions.

ER2014 p.6

BI-EAEA Approach

• The Business Intelligence – Enabled Adaptive Enterprise Architecture (BI-EAEA) Framework comprises:

– Methodology

– Model

– Tool Support

• Model created with URN and tool support in jUCMNav

ER2014 p.7

Why URN

• The User Requirements Notation:

– First international standard to model and analyze requirements with goals and scenarios.

– Goal-oriented Requirement Language (GRL): used to models actors and their intentions.

– Use Case Maps (UCM): used to describe scenarios, processes and architectures.

– KPIs/indicators in URN are used to evaluate performance with respect to some objective or goal.

– jUCMNav is a free Eclipse-based tool that supports GRL and UCM modeling and analysis.

ER2014 p.8

BI-EAEA Approach: Methodology

ER2014 p.9

To Be Scenario

To Be Scenario As Is Scenario

As Is Scenario

Information System

and KPI Modeling

Model Response and

Potential Impact of

Modeled Constructs

Business Goal

Modeling

Evaluate Current

Impacts of Modeled

Constructs

Types of Changes and Enterprise Levels

• What are the types of changes:

– Modifications

– Deletions

– Additions

• Where are changes required in an Enterprise:

– Business Objectives (High Level)

– Decision Makers (Decisions Level)

– Information Systems (IS Level)

ER2014 p.10

BI-EAEA Approach: Sample Model

ER2014 p.11

Hig

h level

Decisio

ns level

IS L

evel

Go

al D

ecision

s In

form

ation

BI-EAEA Approach: Sample Model

ER2014 p.12

Hig

h level

Decisio

ns level

IS L

evel

Go

al D

ecision

s In

form

ation

BI-EAEA Approach: Tool Support

• Analytical Hierarchy Process (AHP) used to get importance and contribution levels.

• Models are checked against OCL rules. 28 rules exist as part of URN profile for Adaptive EA

– An example: “The elements of the Information System must not receive contributions from other actors”

ER2014 p.13

BI-EAEA Approach: Tool Support

ER2014 p.14

BI-EAEA Approach: Tool Support

ER2014 p.15

Response and Potential Impact of Modeled Constructs

ER2014 p.16

Change Need Adaptation Automation

Modification of importance level of a High level,

Decision level or IS level modeled intentional

element.

1) Locate goal, decision or IS.

2) Increase or decrease importance level as required.

3) Check the sum of importance level.

4) Execute As Is Strategy.

2) Pairwise comparison to compute new levels.

3) OCL rule checks violation.

4) jUCMNav evaluations to indicate impact.

Modification of the contribution weights to a High

level, Decision level or IS level modeled

intentional element.

1) Locate contribution.

2) Increase or decrease contribution link.

3) Check the sum of contribution links to goal, decision

or IS.

4) Execute As Is Strategy.

2) Pairwise comparison to compute new values.

Use of jUCMNav contributions overrides for new

weights.

3) OCL rule checks violation.

4) jUCMNav evaluations to indicate impact.

Modification of the KPIs definitions or current

evaluations.

1) Change in KPI values as fed by BI System (or manual

increase or decrease).

2) Execute As Is Strategy.

1) Feeds from BI systems.

2) jUCMNav evaluations (quantitative GRL

algorithm) to indicate impact.

Modification of the desired satisfaction level of a

High level, Decision level or IS level modeled

intentional element.

1) Locate objective, decision or IS.

2) Increase or decrease satisfaction level as required.

3) Execute As Is Strategy.

2) OCL rule checks violation.

3) jUCMNav evaluations (Constraint-Oriented

GRL Algorithm) to indicate impact.

Deletion of a High level, Decision level or IS level

modeled actor or intentional element (their

importance and satisfaction levels as well).

1) Locate actor, goal, decision or IS.

2) Remove actor, goal, decision or IS from model (in a

copy of the model).

3) Check the sum of importance levels of actor’s

intentional elements. Also sum of related destination

contributions links if applicable.

4) Execute As Is Strategy.

3) Pairwise comparison to compute new values.

OCL rule checks violation.

4) jUCMNav evaluations to indicate impact.

Deletion of contribution links to a High level,

Decision level or IS level modeled intentional

element.

1) Locate contribution.

2) Set contribution link to 0.

3) Check the sum of importance level.

4) Execute As Is Strategy.

2) jUCMNav contributions override.

3) OCL rule checks violation.

4) jUCMNav evaluations to indicate impact.

Addition of an actor, or intentional element or their

contributions to a High level, Decision level or IS

level.

1) Include actor, goal, decision, IS (and characteristics) or

contribution link (in a copy of the model).

2) Check that they are linked.

3) Execute As Is Strategy.

2) OCL rule checks violation.

3) jUCMNav evaluations to indicate impact.

Case Study

• Enterprise Architecture of the “Grants and Contribution Program” of a large Government of Canada Department.

• Worked with 4 Enterprise Architects

• The goal model comprised of 4 Diagrams, 8 Actors 40 intentional elements (12 goals, 9 softgoals. 8 tasks and 11 resources), 30 indicators and 102 links.

• A “To Be Scenario” mainly based on a deletion, with the potential impact on the modeled constructs was investigated.

ER2014 p.17

ER2014 p.18

ER2014 p.19

Evaluation

• Work by Yu et al. (2013) identifies 12 characteristics an adaptive Enterprise should have.

• Questions based on the 12 characteristics, were asked and administered to the Enterprise Architects anonymously.

• Framework performance was encouraging.

ER2014 p.20

ER2014 p.21

Evaluation

CHARACTERISTIC

MODEL RESPONSE Weighted

Average (Max: 5) All

(76% - 100%)

Most

(51% - 75%)

Some

(26% - 50%)

Few

(1% - 25%)

None

(0)

Diversity and Variability 3 1 3.75

Uncertainty and Commitment for

known changes 3 1 3.75

Uncertainty and Commitment for

unknown changes 2 2 3.50

Sensing and Effecting Change

(known changes) 1 3 4.25

Sensing and Effecting Change

(unknown changes) 1 2 1 4.00

Barrier to Change 3 1 3.75

Multiple Levels of Dynamics for

documented change 3 1 3.75

Multiple Levels of Dynamics for

ease of use of documented change 4 4.00

Dynamic Systems, Boundaries and

Closure 4 4.00

Actor Autonomy and Alignment 1 2 1 4.00

Business-IT Alignment 1 2 1 4.00

Adaptiveness as a Business

Requirement 1 1 2 3.75

Informal Observations from Stakeholder

• They liked our URN-based modeling approach since:

– It shows the levels where change occurs.

– It offers minimal modeling investments.

– It accommodates changing the granularity of the IS to measure other factors.

– It reflects what they informally do now

– It allows them to disagree sooner!

– The KPIs and satisfactions values accommodates the government’s needs for numbers.

– The models and GRL strategies can be used as documentation trail for analysis and decisions.

ER2014 p.22

Limitations & Future Work Directions

• Limitations:

– What triggers decisions as organizations decide

– Our questionnaires did not have a comparison point

• Future:

– Model and assess enterprise architectures of different sizes and domains

– Look at the cost implication of decisions as adaptation occurs

ER2014 p.23

ER2014 p.24

You’re Invited

http://re15.org/

Thanks! Questions?

Okhaide Akhigbe

okhaide@uottawa.ca

ER2014 p.25