5/2/2011

1

1

Getting Real About Data Virtualization Informatica Data Services

Ash Parikh

May 3, 2011

2

• Makes multiple heterogeneous data

sources appear as one

• Federates data in real-time & also

supports physical materialization to DW

• Abstracts data sources from consumers

and insulates from change

• Hides & handles complexity (quality, rich

transformations, bus-IT collaboration)

• Let’s the business own the data & define

the rules while IT retains control

Source: VMWare, Inc.

Enterprise

Servers

Enterprise

Network

Enterprise

Storage

Hardware

Abstraction

Operating

Systems

HARDWARE VIRTUALIZATION

Logical View of Computing Resources

Portal BI Composite Apps

Enterprise

Data Sources

Data

Abstraction

Logical Data Objects

CUSTOMER ORDER PRODUCT INVOICE

Data

Consumers

DATA VIRTUALIZATION

Logical View of Underlying Data Simple Data

Federation Does Not Cut It

Defining Data Virtualization

5/2/2011

2

3

Here’s Why…

1. Source

2. Target

3. Mapping

4. Workflow

5. Execute/Test

6. Deploy

The Traditional

DI Process

Business

Typical Value Stream Map - Too Much WAIT & WASTE

• The biz gets involved late & does not get what is needed

4

The Problem(s)

5/2/2011

3

5

Business IT

Applications

DW

Unstructured

Data Mainframe

DM DM

Trust

It Takes too Long to Deliver Data the Business Needs!

Change Request … Approve & Prioritize … Analyze & Design … Build … Test … Deploy

6

Business IT

Applications

DW/MDM

Hub

Unstructured

Data Mainframe DM

DM

Spread Marts

DM DM

Trust

Data Is Everywhere & Growing!

5/2/2011

4

7

The

Impact

8

Reports Take too Long

Change

Request

Deploy to

Production

Business IT

3-6 Months

Reporting Scenario: On-going requests for data that is NOT in the DW

?

Change Request … Approve & Prioritize … Analyze & Design … Build … Test … Deploy

Weeks/Days

What if?

•66% of BI requirements change on between a daily and monthly basis

•71% of the respondents said they have to ask data analysts to create custom reports for them

•36% of custom report requests require a custom cube or data mart to answer the request

•77% of respondents cited that it takes between days and months to get their BI requests fulfilled

Source: Forrester Research, “Agile BI: Best Practices for Breaking Through the BI Backlog,” 2010

5/2/2011

5

9

16 Heterogeneous Enterprise Stores With Large Volumes of Data

Different Price Info in Each BU & 1700 Dev Hours to Add 1 Product

Agreement on MEMBER & Attributes Time-Consuming & Painful

Business IT

Product Config Mgmt

(MS SQL Server)

Facets [Benefits, Products]

(Sybase ASE)

Data Warehouse

(DB2)

30,000 Data Marts

(MS Access)

BI

(Cognos)

Portal

(WebSphere)

No Reuse

No Reuse of Data Services for BI, MDM & SOA

SQL Web services

HealthNow Case Study

X

10

Lean Integration &

Data Virtualization

5/2/2011

6

11

Data Virtualization Built on Lean Integration Principles

• Early and on-going business user involvement for agile DI

1. Source

2. Target

3. Mapping

4. Workflow

5. Execute/Test

6. Deploy

1. Logical Data Object

2. Source

3. Integrate - Preview

- Profile at any stage

- Apply rich transformations

- Apply DQ & masking rules on-the-fly

- Federate data without data movement

4. Comment/Tag - Debug

5. Deploy as Reusable Data Services - Web services or SQL

The Traditional

DI Process

The New Agile

DI Process

Business

Business

Original Value Stream Map - Too Much WAIT & WASTE

Optimized Value Stream Map – Cut the WAIT & WASTE

12

Self Service – Analyst Empowerment & Business-IT Collaboration

BI Report

DI Analyst

DI Developer

Data Warehouse

Batch ETL

SQL or Web Service

• Improved analyst & developer productivity

• Easily map sources to physical &

virtual targets

• Quickly find data via integrated

business glossary

• Specify transformations

with reusable expressions

• Include pre-built rules and

mappings (e.g. ETL, DQ)

• Collaborate, test and validate

specification results

• Automatically generate ETL

mappings and SQL views

5/2/2011

7

13

PRODUCT ORDER MEMBER CLAIM

“Virtual Table”

1 week

(vs. 3 months)

Product Config Mgmt

(MS SQL Server)

Facets [Benefits, Products]

(Sybase ASE)

Data Warehouse

(DB2)

30,000 Data Marts

(MS Access)

BI

(Cognos)

Portal

(WebSphere)

REUSE

5x Faster Direct Data Access, Increased Reuse, Improved Governance & Agility

BI Strategy

MDM Strategy

SOA Strategy

14

CRM

Leverage a scalable

& reliable engine

Accounts Involve business early

to define & validate rules

Business

Manager

Analyst,

Steward

Developer

Common

Metadata

CRM Accounts

Access all data,

accelerate profiling Abstract, find, govern

Logical Data Objects

Customer Name Address Category Orders

Support multiple styles

of data processing

Accounts Call Center

ETL

Rich transforms, DQ

& masking rules in RT

DQ

Transformation

Single-click reuse

across all applications

Batch Web Services

1 2

3

4

5

6

7

EII

Optimizations

& Caching

Query

Engine

WS

Server

Case for Making Data Virtualization as part of Your Information Management Best Practices

Informatica Data Services

(Data Virtualization)

5/2/2011

8

15

Next Steps

“SOA Data Integration Architecture

Group”

Forrester IaaS

(Data Services) Wave

Let’s Talk!

Architect to Architect Webinar Series Data Virtualization Architecture & Best

Practices for Agile Data Integration

May 19, 2011

16