Enterprises use IT systems to derive mechanical
advantage through automation of business process steps. Prevalent labor arbitrage outsourcing model has delivered significant
savings in run-the-business IT costs. For a variety of reasons,
change-the-business IT costs continue to be significantly high.
With continually increasing business dynamics and increasing
dependence on IT systems, the problem of managing these costs
with certainty is getting ever more difficult. We propose a model-
driven approach for addressing this problem. The paper discusses
current practice to motivate the problem, describes a model-
driven architecture to address the problem, and discusses various
research, engineering and business challenges for its realization.
4. Enterprise
IT Need 1 IT Need 2 IT Need 3
Transactional IT needs : day-to-day operational
requirements
…
Overview – Problem – State of IT systems
Transformational IT needs : responding to
change; multiple change drivers active- dynamic
supply chains, mergers and acquisitions,
globalization and regulatory
compliances, cloud and mobile technology
Both kinds originate in strategic, tactical, and
operational goals of the enterprise
5. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
6. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Only local context known (if at all);
enterprise-wide context unknown
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
7. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
…
Implement or improve in
Specific
Contexts of Business Silos
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
8. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Locally Operational/Optimal
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
9. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Globally Optimal ?
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
10. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
11. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 2 IT System 3
…
Implementation of Functionalities servicing globally felt need may
overlap due to lack of knowledge about enterprise-wide context
…IT System 1
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
12. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Specific contexts may also result in
Non-interoperable technologies
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
13. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Globally Sub-optimal design of operational
processes
Overlapping Functionalities
Non-interoperable Technologies
…
Transactional IT needs : day-to-day operational
requirements
Overview – Problem – State of IT systems
14. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
…
Headcount-based model/harmonization of
software/ hardware etc. is insufficient
Enterprises are expecting more- outcome-
based pricing- enterprises focus on core
competencies, rest all as an end-to-end
solution by service provider
Overview – Problem- IT Outsourcing
15. Enterprise
1
IT Need 1
IT System 1
Hardware
Infra
IT Need x
IT System x
Technology
Infra
Hardware
Infra
Technology
Infra
IT Plant
Multiple IT Systems of an
Enterprise
Overview – Toward Solution
16. Enterprise
1
IT Need 1
IT
System 1
Hardware
Infra
IT Plant 1
Product Line of IT Plants
Enterprise
2
…
Technology
Infra
IT Need 1
IT
System 1
Hardware
Infra
IT Plant 1
… …
Multiple IT Systems of multiple
enterprises from same domain
Technology
Infra
Overview – Toward Solution
17. Enterprise
IT Need 1 IT Need 2 IT Need 3
IT System 1 IT System 2 IT System 3
…
Transformational needs addressed based on
experts’ knowledge
…
Overview – Problem – State of IT systems
37. Specification & Analysis of
Enterprise Models
Enterprise Simulation
Domain Models
Adaptive Systems
Business Rule
Extraction
IT Plant
Architecture
IT Systems
Architecture
Domain Model
Verification
IT Systems Models for
Change Agility
IT Systems
Architecture
Reconstruction
Enterprise
Models
Domain
Models
IT
Systems/Plant
Models
AnalysisWorld
[Model-driven]
OperationalWorld
[Model-driven]
Research Challenges
38. Enterprise
Analysis World
[Model-driven]
Specification and Analysis of Enterprise
Models
Goal
Purpose Approach
Machine-processable
enterprise models
Enterprise-wide analysis of
satisfaction of properties
Enterprises are complex
interconnected systems of
systems
Use cases- directives,
objectives, policies and
regulatory compliances and so
on, scenarios of interaction
between enterprises-
Outsourcing, M&A, etc.
Early results ontology
based enterprise
models, what-if/if-what
analyses
Techniques for
measuring properties,
quantification
Research Challenges
39. Enterprise
Analysis World
[Model-driven]
Enterprise Simulation
Optimize business process,
supply chains
Reduce risk , cost, improve
planning
Map Simulation Techniques w.r.t their
applicability
System Dynamics Time
Petri Nets workflows
State Machines events
Simulate models in concert
Create mapping between simulation
models and EA models
Create DSL to simulate multiple models
in concert
Capture behavior of Enterprise
Measure and Optimize the
effectiveness of Enterprise goals
by playing out what-if scenarios
Goal
Purpose Approach
Research Challenges
40. Enterprise
Operational World
[Model-driven]
Identify and define formal
modeling mechanism for IT
plant domain.
Identify model requirements
for IT Plant
Identify gaps between the
model requirements and
available Industry standard
models
Identify and define required
modeling elements based on
the assessment.
Pilot it with a suitable domain
Semantics
Traceability
Machine processing form
Multi organization context
Domain [Functional] Models of IT Plant
Goal
Purpose Approach
Research Challenges
41. Enterprise
Operational World
[Model-driven]
Define formal mechanism to
maintain consistency of IT plant
modeling during refinement
Verify the IT plant model for specific
properties
Examine the model refinement
requirement in IT plant
specification
Examine the existing model refine
techniques.
Adopt/enhance the existing
techniques to make them suitable
for IT Plant specification.
For model verification, define
formal mechanism to express
desirable properties of interest
Adopt/extend existing verification
techniques to verify IT plant specs
against desired properties
Consistent model refinement
Verification of desirable properties
of IT Plant specifications
Early detection of model bugs
Domain Model Verification
Goal
Purpose Approach
Research Challenges
42. Enterprise
Operational World
[Model-driven]
Understand and model the
influence of business
products and processes on IT
plant architecture in SP
environment
Map a selected IT plant
landscape in the lines of its
products and processes.
Based on study arrive at the
impact of product and process
changes on the IT plant
architecture.
Improved IT Plant planning
Adaptive technology
architecture
Dynamic provisioning of
resources
IT Plant System Architecture
Goal
Purpose Approach
USP Commoditized
Customer
Facing X √
Back office
√ √
Research Challenges
43. Enterprise
Operational World
[Model-driven]
Goal
Purpose Approach
Investigate mechanisms,
techniques for
(de)composition for variability
capturing
Varying needs of enterprises in
same domain
Customizations per enterprise may
increase app development and
management cost
IT Plant Architecture
Arrive at IT plant architecture
catering to varying needs
Composition and decomposition
mechanisms for IT plant
architecture
Investigate effect of such
mechanisms on non functional
properties
Research Challenges
44. Enterprise
Operational World
[Model-driven]
Use program comprehension
techniques, refactoring, slicing, and
clustering to extract layers and
partitions from the source code.
Using machine learning techniques like
clustering embodied in Lattix tool or
other new techniques.
IT plant Modernization
Most of the architecture/design
documents either outdated or un-
available
Identifying predefined architectural
elements and relations among
them.
Re-construct architecture from the
source code using inputs from
system/application experts.
Architecture Reconstruction
Goal
Purpose Approach
Research Challenges
45. Enterprise
Operational World
[Model-driven]
IT plant Modernization
Extract inherent operational
and business Rules
Identify the constraints and calculation
from the legacy code
Formulate standard and application level
strategies
Apply data dictionary (meanings of
database variables)
Use comments to infer meanings to rules
Apply machine learning algorithms
Rules Representation
Variability
Visualization
Formal representation for verification
Extending SBVR kind of representation
Enterprises are rule driven
Most these are legacy in nature.
To capture ever changing
regulatory policies and regulations
Business Rule Extraction
Goal
Purpose Approach
Research Challenges
46. Enterprise
Operational World
[Model-driven]
Goal
Purpose Approach
Model all aspects of IT systems
with inter-dependencies
Change Impact Analysis
Multi-layered model
Realization thru Model
transformation
Definition of change in terms
of model
Change agility in IT systems
Correctness guarantees
Models IT Systems
Research Challenges
47. Enterprise
Adaptive Enterprise IT Systems-
To meet SLAs-
To reduce people
intervention
Modeling Adaptation for IT Systems
Construct reference architecture for IT
Systems using -
MAPE-k architectural pattern
Feedback Loops
Examine Execution Traces
Investigate data logs and execution traces
to indentify what properties to monitor
and adapt
Language Support
Language to support adaptive software
development
Monitor and adapt the functioning of
IT systems with respect to their non-
functional requirements
Arrive at mechanism to implement
adaptation loop (MAPE)
Operational World
[Model-driven]
Goal
Purpose Approach
Research Challenges
48. Enterprise
How to build enterprise
models ?
Using information
flows?
How to utilize/affect
transformation of IT systems?
Which
abstractions? How
to verify that they
achieve intended
effect?
How to model
business functions?
How to capture
combination of manual
AND automated steps of
business processes?
A language for modeling an
enterprise?
Domain-specific language
drawing on ideas from
event based paradigm,
multi agent systems
Research Challenges
49. Specification & Analysis of
Enterprise Models
Enterprise Simulation
Domain Models
Adaptive Systems
Business Rule
Extraction
IT Plant
Architecture
IT Systems
Architecture
Domain Model
Verification
IT Systems Models for
Change Agility
IT Systems
Architecture
Reconstruction
Enterprise
Models
Domain
Models
IT
Systems/Plant
Models
AnalysisWorld
[Model-driven]
OperationalWorld
[Model-driven]
Summary
Globally Sub-optimal
design of operational
processes
Overlapping
Functionalities
Non-interoperable
Technologies
Dependence
on expert
knowledge
50. IT Plant Architecture
• What
– Arrive at an architecture for IT plant catering to varying needs of customers in
same domain
• Why
– Large enterprises needs plethora of applications to meet their needs.
– Needs of no two customers in same domain are exactly identical
– Various situations Service Providers need to service customer are
• Deliver as-is solutions
• Manage known , unknown configurations
• Manage known , unknown extensions
• Deliver parts and integrate existing systems
– Problems
• Not designed to manage these scenarios
• follow clone-n-own approach
• Increased ADM cost
• Customer not sure of certainty of meeting their needs
51. IT Plant Architecture
• How
– Modularity of IT plant
• Investigate mechanism, techniques for modular design for
– System of system scenario
– Multi-layer architecture
– Capturing Commonality and Variability
– Design time and Runtime variability
– Integration
• Investigate automation for COTS integration
• Evaluation of COTS for quality attributes
– Performance
– Reliability
– Configurabity
– Evaluation of NFR
52. Product Line [Family] of IT Plants
Multiple IT Systems of multiple
enterprises from same domain