Introduction

Master Data Management (MDM) is a discipline for providing consistent content of your key reference data across different parts of your organization.  Examples include:

• Standard customer data
• Standard part data
• Standard pricing data

MDM has emerged in the last several years as a separate enterprise software architecture category as the requirement for consistently defined and maintained enterprise data has become more apparent. 

With the proliferation of different enterprise software systems over the last decade, each with separate data models, yet with data requirements that often overlap, a single system of record for managing an organization’s critical business data has become a practical necessity.  In a study conducted at Merrill Lynch in 2002, the company found six different systems were gathering customer account information, with different addresses and content in each, effectively creating six separate definitions of fundamental account data, and thereby fostering confusion and errors.

Through effective MDM, organizations can eliminate errors, become more efficient in their business activities, and accelerate critical processes such as new product introductions, trade settlement, service provisioning, cross-sell/up-sell, and customer service.  Fundamental business operations such as supply chain planning or inventory costing and valuation, and IT strategies like SOA and BPM, all depend upon the correctness of the data on which they operate.

Analysts have been tracking the master data management industry for several years.  Ray Wang of Forester recently reported:

Data integration, data governance, and training are all major costs, whether companies use in-house or outsourced resources. The study found that, on average, service costs represent almost three times the technology licensing fees. That’s one reason companies are starting with less resource-intensive, registry-style MDM approaches, which have comparatively lower implementation costs.

Master data management can go beyond data integration. MDM generally has a data governance component. Data governance means having to define clear data owners and users, having clear business policies regarding what is good master data and what is not. Master Data Management is also about providing continuous reports and alerts to master data owners and the stewards regarding the health of the master data.

Recent legislation, such as Sarbanes-Oxley in the United States and new international accounting and environmental standards, has driven many enterprises to evaluate their master data management system for the first time. MDM software empowers business users with a best-practice business management process to centralize and directly manage the structure of corporate data.

Technology for Master Data Management

An MDM system is more than a static data warehouse, it is designed for effective updating of information, and for sending updated information to business users and downstream systems.  In addition, an MDM component is the basis for supporting effective information-driven decision support.  Our Information Architecture for Decision Support starts with this layer (called Content Management in those slides).

In the MDM layer, when an update arrives, business logic is run that propagates and applies changes.

This business logic is typically expressed in a scripting language so that it can be easily changed without recompiling the systems.  In Serus’ case, we use the Python scripting language.  The overall architecture is shown below:

Applications of Master Data Management in Semiconductor Operations

Semiconductor manufacturers often lack a global semi-conductor master data update capability, employ multiple, out-of-sync master data management systems, suffer from disjointed manufacturing and inventory management systems and rely on flakey data consolidation systems, spreadsheet systems, report consolidation systems, instead of well defined transaction processing. This results in a lack of operations visibility, operations mistakes, lack of inventory accountability, inaccurate inventory costing and valuation results often based on multiple “shadow” systems lacking traceability, accountability, root cause analysis, and missing key transactional information. 

Dealing with external partners presents additional challenges, including lack of access and ability to process high volume detailed transaction data, systematic business rule enforcement and monitoring, consistent transaction formats, complete lack of automated synchronization and reconciliation, systematically managing exception data, error prone and ad-hoc manual data entry of manufacturing specification into supplier MES systems, no inventory transaction validation against master data specifications, and lack of ability to monitor key bottle-neck processes with real-time feedback.

As engineering groups come out with new products and versions, without a single, fully integrated system to track all manufacturing-related data, it’s difficult for them to ensure the proper components they need are in the pipeline. The more manufacturers have to fall back on e-mail, faxes and EDI transactions to communicate requirements, the more likely they are to have difficulty getting accurate information to the right people at the right time

We apply MDM to the following content:

• Manufacturing specification information
• Part chain information
• WIP transaction reporting.

The benefits of the INCA solution starts with automated high volume detailed transaction processing capabilities now supported by the major semi-conductor suppliers.

• Implementing and managing connectivity with suppliers
• Support for numerous data formats and connection technologies
• Visibility to actual inventory transaction activity

The solution also links product engineering and manufacturing groups, offering:

• Improved collaboration between the two groups
• Manufacturing specification maintenance, with the ability to track frequent design changes
• Routine error correction, to prevent propagation of faulty information to manufacturing execution systems