Upload
others
View
2
Download
0
Embed Size (px)
Citation preview
Software Engineering for Business Information Systems (sebis)
Department of Informatics
Technische Universität München, Germany
wwwmatthes.in.tum.de
Modeling an Information Architecture on
Market Data for Data Governance
Alexander Roschlaub, 07.09.2015 München
1. System Overview
2. Research Questions
3. Research Process
4. Research Artifact I: Requirements
5. Research Artifact II: Information Architecture
6. Data Governance
7. Evaluation
8. Summary
Overview
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 2
System Overview: Visualization
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 3
Data
Warehouse
Market Data
Pool
Market
Data
Portfolio
Data
Client
Vendor
Reporting
BI-Tools
Interfaces
Finance
Data Pool
Portal
Applications
Application Landscape
Data Sources
Portfolio
Data
Market
Data Pool
Data Destinations
Client
System
Clients: Financial & Insurance
Companies
1. Receiving Market Data
2. Outsourcing their financial
reporting
Product:
1. (Enhanced) Market Data
2. Financial report or
analysis
Data Sources:
1. Client’s Investment
Data
2. Vendor’s Market Data
System Overview: Original vs Standard Data
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 4
Positions
Transactions
Investments
Portfolios
Investments
Portfolios
Company
Standardization
Original Database Standard Database
Market Data
Pool
Enrichment Substitution
Problem:
• Disclosure of information about the internal structure of a company
• Large-scale distributed systems with high complexity, little documentation and
no implemented EA
1. What are the requirements (from laws, audits, license agreements, stakeholder)
for processing Market Data?
2. How does an Information Architecture look like, designed for the purpose of
answering requirements on Market Data?
3. What Data Governance measures can be implemented to support the
compliance to requirements on Market Data?
Research Questions
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 5
Research Process
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 6
Identify Problem
Foundation
Requirement Analysis
IA Design
IA Evaluation
Conclusion
Requirements
Literature
Research
Audit
Documentation
Stakeholder
Interviews Concerns
Stakeholder
Interviews
IA
Evaluated IA
Solution Artifacts Sources
[S T 95] S.T. March and G.F. Smith: ”Design and natural science research on information technology”
1. System Overview
2. Research Questions
3. Research Process
4. Research Artifact I: Requirements
5. Research Artifact II: Information Architecture
6. Data Governance
7. Evaluation
8. Summary
Overview
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 7
Examples identified at IDS:
• Solvency II
Quantitative regulations, Governance instructions and Qualitative reporting
• License Agreements
By content, use and source
• Vendor Audits
List of users, Applications and type of data along the lifecycle of data
• Company internal Stakeholder
Data quality, user authorization, need for action and system understanding
Requirements
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 8
Solvency II
Quantitative regulations:
• Rules for assessment of financial assets
Governance instructions:
• Risk control
• Compliance
• Insurance mathematical function
• Internal revision Reporting & disclosure duties
• Qualitative information
Requirements – Solvency II
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 9
[Bund 14] Bundesanstalt für Finanzdienstleistungen (BaFin): “Solvency II: Aufbau und Gesetzgebung”
License Agreements
Content
• Singular products
• Indexlevel and/or Constituents
• Special Data such as sectors, countries
• Frequency: daily, monthly, realtime, EOD, Midday
Data
• Data-, Reporting-, Vendor-, Distribution license
• Within reports, within systems, in raw data format
Use
• Supplied by third-party supplier
• Supplied by different systems
Requirements – License Agreements
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 10
Audits
Data
• type, frequency, storage, access, display, sourcing, exporting
User
• name, authorization, location, external
Hardware
• server, location
Application
• name, owner, supporter, security measures
Requirements – Audits
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 11
Requirements – Company internal stakeholder
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 12
System Overview System Stakeholder
and their Key Concerns
Viewpoints Information Models
Information Architecture
Information Architecture (IA)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 13
[IEEE 00] IEEE Computer Society. “IEEE Std 1471-2000: IEEE Recommended Practice for Architectural Description of Software-Intensive Systems”
[S BU 08] S. Buckl, A. M. Ernst, J. Lankes and Prof. Dr. Matthes: “Enterprise Architecture Pattern Catalog”
• Structural Viewpoint: Displays components, which are processing units of data
• Data Storage Viewpoint: Displays the components, which store data
• Data Flow Viewpoint: Displays the data flow between components
• Infrastructure Viewpoint: Displays the hardware the components utilize
• Authorization Viewpoint: Displays the identifiable objects at each structural
component
• Stakeholder Viewpoint: Displays the interaction and support of the stakeholder
• Governance Viewpoint: Displays, how governance controls can be implemented
and the alignment of such
Information Architecture: Viewpoints
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 14
IA – Structural Viewpoint MDP
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 15
Data
Source
Raw Data
Storage
Data
Storage
Data
Client
Parser
Loader
Data
Catalog Derive
Data
Calculation
Engine
Export
System
File
Archive
Database
Archive
Data
Provision
Legend
Map Symbols
B Component (application or processing unit)
Connection (data or information flow)
Request
Engine
Visualization Rules
Data
Target
A Key Component
Loading Unit
Abstraction
Post-processing
A B Component A is
connected to B
1..n 0..n
storesData Objects
id : String
type : String
frequency : String
Data storage
name : String
Proccessing Unit
name : String
1 1
has
1..n
1..n
connects1..n
1..n
has
Owner
name : String
User
id : String
name : String
access : Boolean
view : Boolean
external : Boolean
Component
name : String
IA – Structural Information Model
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 16
Structural Subviewpoint: Data Flow viewpoint MDP
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 17
Data
Source
Raw Data
Storage
Data
Storage
Data Client
Loading
Unit
Data
Catalog
Post-
processing
Export
System
File Archive Database
Archive
Data
Provision
Legend
Map Symbols
A Component
Data Flow
Request
Engine
Visualization Rules
Data
Target
3.
Retrieving 6. Loading
4.
Archiving
# B Description and Time-triggered, but Event-
dependent Data Flow (= Core Data Flow)
1.
Collecting
2.
Requesting
Archiving
8.
Enhancing
(9.)
Accumulating
7. Filling
10.
Exporting
Providing
C
A B
Time-triggered, Event-dependent
Data Flow from A to B and can be
described by C
B Description and Event-triggered Data
Flow
C
A B
Event-triggered Data Flow from A
to B and can be described by C
5.
Reading
IA – Authorization viewpoints
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 18
Legend
Map Symbols
Identifiable Artifact
Affiliation
Visualization Rules
A Structural Component
MD
Package
MD Object
CD Service
Abbreviations
CD = Client Data
MD = Market Data
CDS = Client Data Storage
CD Object
Enhanced
CD Object
Enhanced
MD Object
MD Source
CD Source Original CDS
MD Storage
Standard CDS
A Structural component
has identifiable object A
B B belongs to A
MD
Request
A B
User
Data Target
Use B uses A A B
MD Fact
Enhanced
MD Fact
1. Data Principles – not yet being utilized
2. Data Quality – the greatest concern of most stakeholders and important for
regulations such as Solvency II
3. Metadata – already heavily utilized by the company, mostly utilized in logging
processes and job triggering
4. Data Access – big concern of the data vendor’s, less so within the company
5. Data Lifecycle – the viewpoint models give an overview of the data lifecycle
Data Governance
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 19
[V Kh 10] V. Khatri and C.V. Brown: “Designing Data Governance”
1. State reconstruction: Contextual information on organizational processes and
services
2. Assessment and measurement
Data analysis: Data understanding and architectural rules
Data quality requirement analysis: identify quality issues and set new data
quality targets
Identification of critical areas
Process modeling: Model of the processes producing and updating data
Measurement of quality: Quality dimensions (completeness, consistency,
currency, volatility and timeliness of data)
3. Improvement: Steps & strategies to reach new data quality targets
Data Governance – Data Quality
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 20
[C Ba 09] C. Batini, C. Cappiello, C. Francalanci and A. Maurino: “Methodologies for Data Quality Assessment and Improvement”
1. System Overview
2. Research Questions
3. Research Process
4. Research Artifact I: Requirements
5. Research Artifact II: Information Architecture
6. Data Governance
7. Evaluation
8. Summary
Overview
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 21
Stakeholder Company Experience
[years]
Work Experience
[years]
Background
External Auditor 5 10 Business
Management
Operation
Manager MDP
4 4 Information
Management
Database
Engineer MDP
7 25 Computer Science
Database
Engineer DW
9 20 Computer Science
Procurement
Manager
4 21 Financing
Evaluation
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 22
Goal:
Proof the usability and information gain, as well as the enabling and
alignment of new tasks through the Information Architecture
Handout of the IA and its elements to the interviewees
Preparation of data material for specific stakeholder (in regard of their concerns)
Interview, which was performed on the basis of four questions
Evaluation: Interview Process
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 23
Interview questions:
1. What is your first impression of the IA in its entirety, after having seen it for the
first time? Is it practicable and understandable?
2. Are the elements of the models correct and the level of detail sufficient enough
3. Are your concerns sufficiently answered?
4. What would you have done different?
Evaluation: Interview Questions
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 24
Stakeholder Answer
External Auditor • High number of components, size of MDP
• Understandable viewpoints
Operation
Manager MDP
• VP easily understandable (IM not so)
• Really good illustration of interaction between client data and
market data
Database
Engineer MDP
• Impressive authorization VP, but too complex
• Simple, but nothing missing
Database
Engineer DW
• Simplicity, well concentrated on core components
Procurement
Manager
• Helpful, simple viewpoints
• Viewpoints not broken up into pieces and enclosed by itself
• Good insight into information demands; Previously missing
requirements
Evaluation: Interview Answers (I)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 25
What is your first impression of the IA in its entirety, after having seen it for the first
time? Is it practicable and understandable?
Stakeholder Answer
External Auditor -
Operation Manager
MDP
• All MDP components illustrated and correct
• No elements missing for MDP – DW interface
• Perfect level of detail
Database Engineer
MDP
• Everything correct and correct level of detail
• Creation of misunderstanding in stakeholder viewpoint
Database Engineer
DW
• All DW components are correct and included
• Good level of detail
Procurement
Manager
• Correct
• Good level of detail
Evaluation: Interview Answers (II)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 26
Are the elements of the models correct and the level of detail sufficient enough?
Stakeholder Answer
External
Auditor
• Vendor requirements sufficiently answered
• Enables stakeholder to gather the required information much
faster through this information basis
Operation
Manager MDP
• Data governance basis for consistent and elaborate monitoring
• Event based viewpoint would have been nice
• Transparency created
Database
Engineer
MDP
• MD interface change could have been addressed more directly,
but answerable
• Interesting and correct user authorization, but too complex
Database
Engineer DW
• Process performance very good answerable
• Regulations much more extensive
• System understanding created
Procurement
Manager
• Can all be answered through authorization viewpoint, but very
complex and laborious to instantiate – not practical
Evaluation: Interview Answers (III)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 27
Are your concerns sufficiently answered?
Stakeholder Answer
External Auditor -
Operation
Manager MDP
More information on application landscape
Use to be evaluated, but governance control desired
Database
Engineer MDP
Minor inconsistencies, but serves its purpose and represents
everything he expected it to
Database
Engineer DW
Interfaces and channels to be better described and event based
viewpoint for market data changes and requirements
Procurement
Manager
Another way to map market data, client data, data source and
data target
Evaluation: Interview Answers (IV)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 28
What would you have done different?
Goal?
Information Architecture for market data processing company
• Requirements on market data
• Data Governance
Carried Out?
1. Literature research for theoretical background
2. Requirement analysis
3. Modeling an Information Architecture
4. Data Governance evaluation
Accomplished?
• Requirements on: Solvency II, Audits, Stakeholder interests
• Viewpoint and respective Information models: Structural, Data storage, Data Flow, Infrastructure, Authorization and Stakeholder
• Environment & means for data quality controls
• Good information basis and illustration of architecture
Summary (I)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 29
Not accomplished?
• Application Landscape not included
• Actual Data Quality Controls for Data Governance
• Not every concern fully answered or answerable
• Sufficient User authorization
Next up?
• More Data Governance controls and Application
• Extend the Viewpoint models
• Automatization
Summary (II)
© sebis 150907 Roschlaub: Modeling an IA on Marekt Data for Data Governance 30
Technische Universität München
Department of Informatics
Chair of Software Engineering for
Business Information Systems
Boltzmannstraße 3
85748 Garching bei München
Tel +49.89.289.
Fax +49.89.289.17136
wwwmatthes.in.tum.de
Alexander Roschlaub
Student
-
Any Questions?