Rather than force-fit data into a central repository, a more federated approach is emerging under the auspices of master data management (MDM). A combination of methodology and technology, MDM establishes practices that foster consistency and accuracy across all systems and divisions, while identifying and managing interrelationships of core data.
Because all the information is collected from the original source and then made available instantaneously, the redundancy of reentry is removed. Eliminating reentry speeds processes and reduces errors.
Key to the MDM principle is the creation of “master” data — in which a central application ultimately owns data common to all facets of the organization. Additionally, master data links them to other subdefinitions held by other lines of business.
As the central system of record, the master data becomes the building block for allowing multitudes of applications — such as SRM, Legacy system, call center linked to a CRM, and Web channel systems — to procure one version of truth from the MDM system.
So Are there any pitfalls in so ensued consultant approach.
The federated approach to CDI has gotten some traction as a new way to tackle the problem. In this approach, customer data is not persisted in a physical repository but accessed and aggregated on the fly and presented to users as a series of ‘views’ into dispersed customer data( IBM tool Centric). On the positive side, it avoids political battles around data ownership as itleaves data where it lies and accesses and aggregatesit dynamically. On the downside:
• Does not address the issue of inaccurate,redundant, or inconsistent master data; it merely
puts a favorable virtual view on top of the data to serve the needs of that particular instant. Data quality is only as good as the federated data sources.
• Designed for access only. If many systems have to be updated to complete a transaction, the federated solution will have to build complex handshaking and roll-back logic to be viable.
• In a recent study I read that it is only suitable for small-scale information movements as it has to access many systems in real-time and perform in-memory aggregations,filtering, and transformation on data that may not be optimized to be viewed in the manner being requested.
• Information management functionality such as versioning, rollback, audit control, or validation is handled by the individual systems and can be inconsistent or inadequate. Creating and validating data relationships for data stored across multiple systems is not possible.
So having listed some drawbacks of MDM what keeps it going ?
Such an approach is technically more complicate, but if you use a technology that is capable of being federated (which these do) then the overhead is not great, but you have the key advantage of retaining global v local flexibility.
- Core master data have common definitions and governance
- Makes project viable
•Diffuses organizational conflicts
•Supports evolutionary approach
- Resilient to change
- Highly Recommended
- Conceptually complex
- Hard to define scope