Implementing MDM for BI & Data Integration by Kabir Makhija.

Implementing MDM for BI & Data Integration by Kabir

Makhija

What’s the holdup?

What is Master Data?Any enterprise has 6 mutually exclusive, collectively exhaustive (MECE) types of organizational data, which are:

Type 1) Transaction Structure Data – Dimensional context to business transactions. Eg: Products, Customers, Departments etc.

Type 2) Enterprise Structure Data - Inter-relationships between organization elements. Eg: Chart of Accounts, Org Structure, Bill of materials, etc.

Type 3) Reference Data - Set of codes, typically name-value pairs that drives business rules. Eg: Region Codes, Customer Types etc.

Type 4) Transaction Activity Data - These are the transactions themselves. Eg: Purchase Order data, Sales Invoice data etc.

Type 5) Metadata – Data about Data

Type 6) Audit Data – For Compliance

Typical understandi

ng of “Master Data”Holistic view of “Master Data”

Comprehensive view of “Master Data” encompasses Transaction Structure Data, Enterprise Structure Data and Reference Data.

Master Data Management

Master data management (MDM) enables dependable cross-system, enterprise-wide business processes and analytics – ensuring that everyone involved in the process has access to the same information and knowledge

MDM is the opportunity to: •Implement a data integration platform that can access the facts about core business entities from anywhere in the enterprise

•Automate the creation of a single logically correct view, based on business rules that agrees with the facts in the real world

•Deliver that high quality master data to the current suite of business applications in real time

Customer FAQ

• What is MDM ?

• How to get started ?

• Who are the vendors ?

• How do the products compare ?

• What is the ROI ?

System Integrator FAQ

• Does the organization consider data governance as a nice to have or must have ?

• How does the client rate the current Data Quality ?

• What is the current solution in place ?• Is an enterprise data model available ?• Operational and / or Analytical MDM ?• Is there a service oriented architecture

?

Business Drivers

• Runaway Costs

• Missed Revenue Opportunity

• M & A Integration

• Support existing initiatives

• Regulatory Pressures

MDM is at the heart of business decisioning – Needs “Total Alignment” with corporate vision

Organization should be geared for “Change” – Cultural issue

Keeps Evolving over time - MDM systems are dynamic in nature

MDM is not just Technology – Process Institutionalization is critical

MDM – Critical Leverage Points

Data Management & Governance is crucial for Business Buy-in

Challenges Vs Solutions

MDM Vendor Offerings• CDI / PIM• DQ• ETL

Enterprise Challenges• Scoping • Data Governance• Organization Culture• Prioritization

Enterprise Solutions• Enterprise Data Warehouse (EDW)• Data Federation• Customer Relationship Management

(CRM)• Enterprise Resource Planning (ERP)

Domain Specific Solutions• Customer Data Integration (CDI)• Product Information Management (PIM)

Technology Solutions

Implementation Styles

Single Physical Data Store Approach– Single consolidated master data store that contains master data from

multiple source systems– Latency depends on whether batch or on-line data consolidation is

used, and update frequency

Federated Approach – Virtual business view of the reference data in source systems is

defined. Used by business applications to access current master information

– May employ a metadata reference file to connect related master information together based on a common key

Hybrid Approach– Combines data consolidation and data federation approach– Common master data (name, address, etc) could be consolidated in a

single store, but master data unique to a specific source application (customer orders, for example) could be federated.

– This hybrid approach can be extended further using data propagation

Customer Data Integration

Multiple & Federated Data Sources

•Standardization of different sources that store data in different formats

•Integration of data from multiple data sources

•Consolidating diverse data integration tools

•Global time synchronization in multi-geography systems

•Identification of common batch windows for extraction and processing

Data Cleanliness & De-Duplication

•Conversion from free form text of Source systems

•Cross-organization data standardization

•Geo-coding and cleansing

•Consumer data de-duplication

- Identify a customer uniquely across organization - Identify the parameters for house-holding - Defining survivor and merge rules

Product Information ManagementData Source / Domain

•Standardization of data source format and layout across the multiple regional databases

•Consistency of data type and allowable range of values

•Seamless handling of changes to data attributes

•Reusable framework for implementing new data sources and regional databases

Data Completeness & Validity

Data Lifecycle Management Data Management

•Master data completely updated with all regional data

•Checks and balances to ensure that source - regional data and regional- master data match

•Master database is maintained with integral data

•Setup and Maintenance of Validated meta data

•Maintenance of obsolete data from source system in the master

•Tracking and handling of bulk movement of data between departments

•Efficient historic treatment of changing data

•Optimal storage mechanisms and capacity planning for regional and master databases

•Efficient data roll-up decisions for SKU realignment- SKU to department or brand can be automatically realigned & SKU orients under the brand

•Reduced Data mismatch with respect to SKU realignment

BI Solution Architecture

Deployment Framework

Steps in a MDM Implementation Identify sources of master data

Identify the producers and consumers of the master data

Collect and analyze metadata about for your master data

Appoint data stewards

Implement a data-governance program and data-governance council

Develop the master-data model

Choose a toolset

Design the infrastructure

Generate and test the master data

Modify the producing and consuming systems

Implement the maintenance processes

MDM Maturity Levels

Level 1 • Data Integration with minimal focus on DQ

Level 2 • Managing basic Data Quality

Level 3 • Master Data within Silos

Level 4 • Enterprise Master Data for a single domain

Level 5 • Cross-Enterprise Master Data for multiple

domains

Marketplace

Mega Vendors Pure Play Data Quality

IBM Siperian DataFlux

Oracle Kalido Trillium

SAP Initiate Informatica

Microsoft Purisma Netrics

Teradata, TIBCO, etc OneData,Acxiom, etc

Pitney Bowes, Zoomix, etc

Thank You Research Credits : Hexaware BI & A team