Operational Data Store

Currently, we are building for a preeminent Colorado child sponsorship organization an Operational Data Store that fits into a larger picture which introduces a master data management system and an enterprise data warehouse.

The implementation of the ODS is especially challenging since some future avenues with respect to the final purpose of the data store must be kept open and the integration of other off the shelf compents will be explored.

The co-founders of the Data Warehouse concept, Ralph Kimball and W.H. Inmon represent significant differences in the design and architecture of the enterprise data warehouse. Inmon advocated a “Dependent Data Mart Structure” whereas Kimball advocated the “Data Warehouse Bus Structure”.  While one finds in this domain two main approaches (and evolutions of them) based on an underlying concept, the situation for the Operational Data Store is even muddier.

About the only agreement is a definition that characterizes the ODS as a "subject oriented, integrated, volatile, current value oriented data store containing only corporate detailed data". At the same time, due to the fact that data warehouses are being used for data mining via analysis services, the operational data store is becomming increasingly integrated with the data warehouse to begin with and can be viewed as a very responsive real time data warehouse.

The following table is a collection of characteristics that we consider as differentiating between an OLTP, ODS, and EDW:

Characteristic
Audience Data Access Data Content
Data Granularity Data Organization Data Quality
Data Redundancey Data Stability Data Update
Data Usage Data Size Data Structure Stability
Development Methodology Operational Priorities Philosophy
Predictability Response Time Return Set

 

 We found that we can quickly align the customers' thinking by emphasizing initially on the following:

  • Current Value only
  • Time horizon: 60-90 days
  • Operations: insert, update, delete, change, read, replace
  • Query size yields small amount of data

Examples

  • OLTP query: Is the child currently sponsored
  • ODS query: How much did a sponsor donate during the last 90 days
  • DWH query: Has the sponsorship risen during the last 5 years for the West Africa Region