Oops! The input is malformed!
Originally published March 14, 2013
In order to enable agile analytic development in a modern analytic competitor, a support infrastructure needs to be put in place. Giving data the best chance to succeed by putting it in the best store for its intended workload means the architecture must be enabled with the tools necessary to move data, access data across platform and leverage common data. This means data integration, data virtualization, master data management, data governance and program governance. These are essential tools for the analytic architecture.
Data integration not only moves data, but also it leaves data behind in the “source” and therefore creates redundancy. However, some level of data redundancy has long been accepted by analytic teams.
This is most evident as a means to load data warehouses and data marts. However, there is much more movement necessary in an analytic architecture. Moving data from source to target is part of the continuous cycle within the analytic architecture. It’s both built into the daily cycle and needs to be present for immediate needs.
Focusing on a primary means of data integration, and establishing a center of excellence for it, facilitates the development of the analytic architecture by centralizing best practices, tool support and vendor culture while facilitating the use of data integration in the enterprise.
Data virtualization is the ability to incorporate data from multiple platforms in a single query. Similar to data integration, there are going to be workloads that utilize data virtualization daily and there are going to be workloads that need data virtualization immediately. If you are stepping up to being an analytic competitor and developing the analytic architecture, you will have numerous vessels to put data into today. This will result in the need for queries across technologies. Data virtualization is the technology supporting this need.
We not only want to empower the analytic ecosystem with tools, but also we want to empower it with data. Every application needs master data. How many don’t need a customer list? Or a product list? Since they mostly do – and they will all certainly benefit from working off of the same list – master data management takes this work off the table for each project and centralizes it.
Master data management is a discipline unto itself. There are not only the integration aspects, but also there is, very importantly, the data origination aspects to the master data. Taking the significant work of master data origination off the table and integrating with a source built for integration is part of an analytic architecture.
Finally, we have something that is “light” on technology but heavy on softer skills. Data governance is the principle-forming body of the organization that gives guidance, at an appropriate level, to the formation of the analytic architecture. Data governors will contribute to data quality, data origination, and data transformation rules and provide unique insights to the build team in terms of the longer-term technical needs that the analytic architecture must satisfy.
Program governance is also an organizational body, but directs higher-level decisions such as giving priority or funding to competing projects. Program governance is able to uniquely understand the subtleties of practically keeping a program moving forward. Its participants are also key in spreading desired messages throughout the business community to attain and keep strong support for the information management and analytics programs.If you embrace the idea that a modern organization’s information management needs are going to be robust and diversified, then you need to facilitate the cost of entry for necessary new data stores into the organization. Each one of these enablers will be needed, or will save the organization measurable time in delivering a new data store for the analytic architecture and integrating it with the rest of the organization.
Recent articles by William McKnight