1
Informatica Data Federation Overview
Claudia ChandraProduct Management
2Informatica Confidential and Proprietary
Agenda
• Why Data Federation?
• How It Works
• Customer Use Case Scenarios
• Demo
3Informatica Confidential and Proprietary
Enterprise Data IntegrationPhysical data integration
ETL (Extract-Transform-Load) Approach• Extract data from source systems• Integrate, clean, organize, aggregate data via transformation• Store data permanently into target system (usually RDBMS)• Client consumes data from target system (usually via SQL)
Strengths
Holistic Information
Enterprise Platform
Lifecycle Productivity
SOA
Benefits• Data cleansing, rich transformations• Historical data, deterministic results
• Model-driven architecture• Reusable components
• Robust & hardened for large volumes• Protects operational systems
• Collaborative development
4Informatica Confidential and Proprietary
Complementary ApproachData federation (EII)
Approach: provides a virtual layer for defining virtual views across multiple data sources for read and write access by applications
• Query data from source systems• Integrate data• Deliver to client via access protocol determined by client (SQL, Web Services,
Java, etc.)
Strengths
Holistic Information
Enterprise Platform
Lifecycle Productivity
SOA
Benefits• Current data delivered on-demand• Unconstrained access
• Rapid development
• Optimized query performance and caching
• Reusable data views abstracted from data sources• Web Services-based data access
5Informatica Confidential and Proprietary
Extending Enterprise Data IntegrationData Federation + Physical Data Integration
• Holistic Information• Historical and Current Data
• Cost-Effectively Scale• Greater Flexibility to Optimize Data Architecture• Physically Move Data Based on Proven Demand
• Reduce Time to Results• Faster Response to Business Users• Prototype Virtually, Lock in Requirements, Implement Physically
For many use cases, extending PowerCenter with Data Federation
yields significant benefits
6
Data Federation How It Works
7Informatica Confidential and Proprietary
Data Source 1 Data Source 2
Step 1: Build the ViewStep 1: Build the View(A ViewView contains MetadataMetadata on the actual data source – format, type, location, access, etc.)
Data Federation Option– How It WorksCreate View
Data Federation Option View
8Informatica Confidential and Proprietary
Data Federation Option View
Data Source 1 Data Source 2
1 A B
2 C D
3 E F
4 G H
1 a b
2 c d
3 e f
4 g h
1 A B a b
2 C D c d
3 E F e f
4 G H g h
Step 2: Aggregate Data(Queries are optimized for speed and efficiency)
Step 3: Deliver DataStep 3: Deliver Data(Combined data creates meaningful information)
Data Federation Option– How It WorksAggregate & Deliver Data
Step 1: Build the View(A View contains Metadata on the actual data source – format, type, location, access, etc.)
Portals BI ToolsWeb Services
9Informatica Confidential and Proprietary
PowerCenter Data Federation Option
Business User
Disparate data sources
Issue multiple parallel subqueries
Invoke view; validate security
Generate alternative query plans; pick the best
Format results and return data to client systems
Join, map, and/or aggregate results
Retrieve results from data sources
Display results to userUser requests information through application
Data Federation Option– How It Works
Data Warehouse
RDBMS
XMLPackaged Applications
Web Service
10
Data Federation and PowerCenterUse Case Scenarios
11Informatica Confidential and Proprietary
Data Warehouse ExtensionSupplementing with New Data Sources
Data Warehouse
Operational Data
Application
Application
Database
Applications, Portal, BI Tool, etc.
VirtualView
ODBC / JDBC / SOAP
Message Queue
Data Federation
PowerCenter
Real-time data
Approach • Physical: move data
into data warehouse• Federation: combine
information from data warehouse with operational data and data from message queue
Benefit • Accelerate addition of
new data sources to data warehouse
• Combine historical and current data
12Informatica Confidential and Proprietary
Next Generation Data Warehouse Prototype and Build
Relational DatabasesFiles
MetadataRepository
Applications, Portal, BI Tool, etc.
Data Warehouse
VirtualView
Data Federation
Power Center
Approach • Federation: create
prototype with virtual views, validate with users
• Physical: with requirements locked in, build robust data integration workflows to create physical data store
Benefit • Accelerate development• Ensure alignment with
user requirements
ODBC / JDBC / SOAP
13Informatica Confidential and Proprietary
Next Generation Data MigrationPrototype, Migrate and Insulate
MetadataRepository
Applications, Portal, BI Tool, etc.
Physical Target
Virtual data layer
Data Federation
Approach • Federation: prototype
migration with virtual target, validate
• Physical: migrate the data from the legacy system into the new physical target
• Federation: insulate downstream systems from change
Benefit • Accelerate migration• Make migrations
transparent Legacy
PowerCenter
ODBC / JDBC / SOAP
14Informatica Confidential and Proprietary
Data ServicesVirtualization of data sources
Applications, Portal, BI Tool, etc.
Virtual data layer
Data Federation
Approach • Physical: create
physical data stores• Federation: create an
abstraction layer enabling virtual, federated access to data in back-end systems
Benefit • Protect consuming
systems from details of underlying implementation, potential changes
ODBC / JDBC / SOAP
Data Warehouse
Operational Data Store
Application
Application
Database PowerCenter
Application Data
15Informatica Confidential and Proprietary
Informatica Differentiators
• Integrated metadata and engine• Reuse logic between data federation and ETL (bi-directional)• Rich ETL transforms—cleansing, etc.
• Universal connectivity• Make heterogeneous data sources available to data federation
• Enterprise performance and scalability• Federated queries on grid
• Metadata-aware data federation• Data lineage • Profiling “metadata”
16
Thank you