Карта сайта
Версия для печати

Преимущества многогранного подхода к MDM-стратегии

1 августа 2013 Согласно компании Gartner, к концу 2014 года около 70% компаний из списка Fortune 1000 планируют завершить не менее 2 проектов по внедрению решений, предназначенных для управления мастер-данными. Таким образом компании снизят затраты на инициативы интеграции и анализа данных благодаря многогранному подходу. Каким образом? Читайте статью на нашем сайте. (Материал опубликован на английском языке)
In financing and implementing these single-domain MDM platforms – which usually focus on either product or customer data – organizations will have nearly doubled the cost and resources associated with the purchase and implementation of one multi-domain MDM solution.

They will have also significantly reduced their chances for enterprise-wide data integration, while succeeding in complicating issues of governance, data quality, and data enrichment – all points that MDM is supposed to enhance, not exacerbate.

Most importantly, according to Charlie Lawhorn, Senior Vice President of Stibo Systems, they will have reinforced a silo-based culture of Data Management that misses the entire point of MDM:  "The premise of MDM is about trusting and sharing. The whole goal is to get this collection of data accessible to all of those people and systems that need it, so that they can trust it and it can be shared. If you’re just logging those things in within silos and departments that are within the organization, you’re not really solving MDM.”

Single Domains vs. Multiple

A 2012 survey from Forrester maintains that only 9 percent of MDM professionals are utilizing multiple domain MDM solutions, although nearly half of them asserted that they have at least three domains requiring MDM. These figures partially reflect the single-domain origin of MDM technology, as well as the following boons:
  • Domain specificity – single-domain solutions have pre-template data models for either customer or product domains, and a number of highly domain specific adaptors including aspects of customer acquisition, data cleansing, product attribute standardization, and flexible model support.
  • Industry specificity – certain MDM solutions are geared towards particular industries, such as health care.
  • Familiarity – Employees in the particular department that a domain-specific MDM is geared towards have a degree of confidence about its reliability and dependability since they use it most.
It is not uncommon for organizations using single-domain MDM solutions to have more than one hub, or more than one solution to address different domains. As beneficial as this option is for those with access to that particular silo, complications arising from data governance, which are worsened by potential Big Data initiatives, have the potential to offset them.

Supports Growth

Multi-domain MDM platforms, in contrast, can provide a panacea for virtually all of the complexities of single-domain solutions. This fact is most often seen in terms of cost, since organizations can purchase one solution that not only assists with the traditional domains of customer and product data, but also with the myriad others that include supply, location, assets, employees, and more.

Also, no matter what the size of an organization, it is rarely easy to stratify data needs into single domains. Marketers, for instance, who may need customer data, will also need access to product data so that they can pitch and sell accordingly. According to Dan Power of Hub Designs, problems associated with MDM frequently occur across domains so that what begins as a point solution for a particular domain evolves to include aspects of others. This trend is magnified as an organization grows and its MDM needs expand beyond a department to encompass the entire organization. Power noted:

“That leap from department to enterprise is very difficult when you’re talking about Master Data Management. It’s difficult to get sponsorship and funding; it’s difficult to explain why you have to go out and buy a whole new platform again after you’ve just implemented a platform.”


Governance Implications

One of the most salient wins for multi-domain MDM platforms is in Data Governance. With single-domain silo-based solutions, it is difficult for Governance Councils to ensure that standards are being met. End users tend to have a great deal of autonomy in updating and storing data, which may result in multiple definitions of terms that ideally should be concurrent throughout the entire organization.

With multi-domain MDM solutions, it is much easier to ensure that standards are being met since there is only one repository for the Master Data. Users will have substantially less autonomy in creating definitions and rules for data, since the overall architecture will be simplified resulting in greater transparency. As a result of better principles of governance, cross-functional collaborations between departments is more likely to occur, resulting in increased efficiency and better resource allocation.

One of the chief jobs of those involved with Data Governance is assuring data quality, enrichment, and a level of trust that might not otherwise be there. The job for Data Stewards and Governance Councils is substantially easier when there is only one multi-domain MDM platform that has full data integration. Quality programs only need to cleanse data once, and there is less likelihood of redundant data and other factors that result in distrust. Power discussed the effect of multi-domain MDM on data governance.

“The multi-domain aspect ensures that you’re bringing together different parts of the company, different groups of data from different parts of the company, and that you’re governing the end result. You’re not just throwing all of this into a hub and hoping for the best. There are business processes, there are rules, there are policies, and that’s being supported by senior management so that people now have a trusted repository of data.”

Big Data

The growing movement towards Big Data is functioning as a driver towards multi-domain MDM. The variety of data that Big Data provides can apply to any number of domains. More importantly, in order to derive any sort of meaning from Big Data, an organization needs to sufficiently master its own proprietorial structured data.

It is more difficult to integrate Big Data into single-domain MDM for the simple fact that the variation of the data may not apply to that particular domain – although it might for others. Without a multi-domain MDM that can offer an entire view of the relationship between Big Data and a company’s proprietorial data, the value of Big Data becomes reduced. According to Power:

“People are very taken with the idea of Big Data and Enterprise Information Management, so they’re looking at an even bigger picture beyond just Master Data. It makes no sense for them to do a single domain solution at this point in the game in 2013. Nowadays they’re looking at ‘how do I handle Big Data challenges like social media and unstructured data’ since 80 percent of the information generated every year is unstructured.”


Multi-Domain Must-Haves

There are plenty of multi-domain MDM platforms available, each of which has particular strengths and weaknesses – not the least of which revolves around the amount of customer support a vendor is willing to provide. The following checklist denotes a series of necessary capabilities that makes a particular product worth purchasing, and illustrates several valuable features of this technology.
  • Data quality: Systems should have automated data profiling that expediently examines data quality levels; incoming records should be standardized and cleansed according to user-provided business rules; tools should verify contact information for accuracy.
  • Integration: The product should include APIs for real-time, batch integration and Data-as-a Service.
  • Governance support: Competitive platforms include a module specifically for governance which is focused on the usage of data stewards; the solution’s individual components should unite in the user interface to provide an integrated view.
  • Lineage: Multi-domain MDMs should preserve the raw data initially received in the hub and provide a trail for auditing golden records as they are changed.
  • Modeling: Solutions should enable flexibility so that the platform behavior changes according to data modeling changes; the hub should represent relationships and entities found in the data.
  • Loading: Platforms should enable data migration and updates that prevent duplicates within the incoming data and that found in the hub.
Future Adoption

At this point, there are advantages for utilizing both single and multi-domain MDM platforms. The primary difference is that a growing number of multi-domain MDMs can provide much of the functionality of the former, as well as much more. As organizations grow, it may be to their advantage to abandon the traditional silo-based approach and fully integrate their MDM enterprise-wide. With both issues of governance and the relevance of Big Data operating as drivers, the trend towards multi-domain MDMs should only increase.


Source:  dataversity.net