This presentation was given at JUDCon 2013, Jan 17,18 at Bangalore. Presented by Prajod Vettiyattil and Gnanaguru Sattanathan. The presentation deals with the Why, What and How of Data Services and Data Services Platforms. It also explains the features of the JBoss Enterprise Data Services Platform.
The need for Data Services is explained with 3 Business use cases:
1. Post purchase customer experience improvement for an Auto manufacturer
2. Enterprise Data Access Layer
3. Data Services for Regulatory Reporting requirements like Dodd Frank
2. Enabling Data as a Service
with
JBoss Data Services
Prajod Vettiyattil Gnanaguru Sattanathan
Twitter: @prajods Twitter:@gnanagurus
Website: bushorn.com
2
3. What this session is about
The why and what of data services
How data services work
Use cases
JBoss Data Services Platform
3
5. Proliferation of data
Data Consumers
Custom Employee
ERP CRM Accounting Billing
er portal portal
Partner Vendor
Finance Marketing Sales Management Management
Content
Mainfra Manage
SQL File NoSQL Email ERP
me ment
System
Data Sources and Data Managers
5
6. Proliferation: so what ?
• Multiplicity of connections
– High development cost
– Huge operational overhead
– Difficult and risky to change Data Sources/Managers
• Dispersed data connectors
• Data duplication
– Too much ETL
– Lines of Business copies data
• Duplicated data aggregation
• Impossible to create “Single source of truth”
• Data ownership issues
• No comprehensive view
– No data movement dashboards
– Location of data and its status
6
8. Data Services and DSP
The basic view
• DSP = Data Services Platform • Presents the data as a service to the
• Abstracts the data consumer
managers/sources • ETL++
C1 C2 C3 Data Consumers C4 C5 C6
Data
Service 1
Data
Service 2 Data Services Platform Data
Service 3
Data
Service 4
D1 D2 D3 Data Managers D4 D5 D6
8
9. Dashboard in a DSP
Data
Data
movement Errors
Connections
status
Data Dashboard
Error
Failures Alerts
Corrections
9
11. Features of a DSP
• Enables architecture principles
– Separation of concerns
– Protected variations
• Data adapters
• Data mapping tools and standards
• Data caching
– Local and distributed
• Service search and reuse
• Data security and data usage audit
• Data access control
• Central channel for all data requirements
• Data dashboard
• Configurable performance and reliability
11
13. Auto manufacturing supply chain:
Requirements
• Vehicle ownership experience
• Business Process Automation
• Disparate data sources
• Multiple data feeds
– Parts catalog
– Prices
• Dealer updates
– Parts consumed
– Parts replaced
– Part failure statistics
• Customer feedback
– Post purchase
– Breakdown support
– Service Quality Dashboards
• Integration solutions based on batch transfers
– Unreliable
– Not traceable
13
14. Auto manufacturing supply chain:
Layer Diagram
Customer Business Activity Monitoring
Experience
Dashboards Business Processes
Enterprise Service Bus
Data Services Platform
Parts
Breakdown Customer Parts Dealer Dealer
supplier Customer
reports feedback Catalog feeds
feeds Info Master
14
15. Enterprise Data Access Layer:
Requirements
• Golden copy / System of Record / Single source of
truth
• Shared services team for Enterprise Data
Management
• Data usage audit
• Data access control
• Reduce request load on Data Management team
• Reduce data maintenance costs
15
16. Enterprise Data Access Layer:
Layer Diagram
Enterprise Data Consumers
Data Services
Virtual DB Data Services Platform Metadata
Data base Data Access
Auditing Data Aggregation
drivers Control
Content
Partner
Mainframe Management Partner Customer Employee
Data
System Info Master Info
16
17. Reporting risk for securities:
Requirements
• Internal and external reporting
– Risk and margin
• Centralized risk capture and management
• Calculate risk from different customer activities
• Report consolidated data to comply with regulation
– Dodd Frank
– Sarbanes Oxley Act (SOX)
• Dashboards for higher management
17
18. Reporting risk for securities:
Architecture without DSP
COTS Trading Customer Government Reporting
Partner Apps
Systems facing Apps Systems Applications
Execution Liquidity Position
Order Mgmt Order Book
Mgmt Mgmt Mgmt
Price feeds
Enterprise Middleware Systems Trade feeds
(MQ, ESB, FTP, File shares)
Trade Ref Data Payment
Margin Mgmt Clearing
Matching Feeds Systems
Custom built Risk Ref Data
Settlement Accounting
Apps Management Mgmt
18
19. Reporting risk for securities:
Patterns in this requirement
• Regulatory requirement for transparency
– Cannot be met by opaque internal systems
• Data Sources
– Large number of them
– Internal and external
• Reports are read heavy
• No real time data requirements
– once a quarter or once a year
• No excuses for incorrect data in reports
• Non-discretionary spending
19
21. Architecture
Data consumers
(Custom Applications, COTS products, Business Processes, Business Services )
Data interfaces
• The EDS platform (JCR API, Web service, JDBC, ODBC, OData,..) • Parts of the architecture
– v5 Runs on SOA-P – Data interfaces
• Teiid Metadata – Data adapters
Data virtualization –
• ModeShape repository Data virtualization
– Metadata repository
Data Adapters
SOA Platform
Data Services Platform
SAP Sybase Flat file XML SalesFo Oracle Cassan Mongo
rce dra DB
Data Sources
21
22. Data sources
Oracle IBM MS MySQL PostgreS Sybase
DB DB2 SQL QL
Server
Greenpl Teradata Netezza Ingres Mondria MetaMa
um n trix
LDAP Salesfor Delimite XML Web Apache
ce d file file services Hive
MS MS JBoss JBoss TIBCO IBM
Excel Access Messagi HornetQ MQ
ng
22
23. Data Mapping
• Teiid Designer
– Map actual data tables using transforms to virtual
tables
– MDD; use Data Models, not SQL
– Semantic mapping
– Virtual procedures
• A set of SQL statements, similar to DB stored procedures
23
24. Data Standards
• JCR
– Java Content Repository(JSR-283)
• OData
– Open Data Protocol
• JDBC
• ODBC
• Others
– S-RAMP
– An SOA repository spec, OASIS
– Web Services
– REST
– JMS
24
25. Access control and Audit
• Teiid
– passwords
– MembeshipDomains for authentication
– Data roles
• Fine grained access and visibility control of tables
– CRUD level permissions for VDB
– LDAP integration
• ModeShape
– LoginContext
– AuthenticationProvider
– Role to Action mapping
25
26. Teiid and ModeShape
Data type Teiid ModeShape
Approach Relational Hierarchical
Metadata repository Not suitable Yes
Content repository Not suitable Yes
ACID transactions Yes Yes
SQL queries Yes Yes(JCR-SQL)
Flat file data source Yes Not suitable
Relational DB data source Yes Not suitable
Schema Fixed Optional
NoSQL data sources Not suitable Yes
Stores data No Yes
26
27. Summary
• Data Services
– Why
– What
– How
• Use cases
– Auto Manufacturer
– Enterprise Data Access Layer
– Regulatory Reporting
• JBoss DSP
– Data virtualization
– Teiid
– ModeShape
27
28. Questions
Prajod Vettiyattil Gnanaguru Sattanathan
Twitter: @prajods Twitter: @gnanagurus
Website: bushorn.com
Our Open Source Middleware Group on LinkedIn
http://tinyurl.com/be6e93q
28