Designing an operational data store is probably one of the most critical set of decisions that an architectural team will undertake in the course of rolling out an enterprise level data management program. The considerations for the architectural design of an operational data store should strive to encompass the major drivers of demand for the ODS. The major considerations are primarily driven by the specific requirements of the ODS, and the requirements need to be crystallized well before the design planning sessions are undertaken. Of course, some or many of these considerations might not be entirely known, but it is imperative that a joint task force of the business and technology teams is set in place that ttempts to set as many of these considerations in stone prior to design sessions.
The important design characteristics of the ODS include
- Expected Data Volumes
- Data Latency/Recency
- Data Quality
- Flashback/Replay Options
- Change Data Capture
- Data Corrections
- Data Reconciliations
- Load Process Suspension
- Suspense Data Processing
- Data Model Style (Normalization Levels)
- Integration with external data services
- Data Publication Mechanism
This list, of course, can be modified to add more characteristics as need be. However, this list is a good starting point for the design process to start. Both sides of the team, technical and business, need to be involved in discussions around each of these characteristics, since the overall design and architecture of the ODS is driven by where the specific requirements are for each of the items on the list.
Details about each of these characteristics will be posted to Infonitive over the next few days as we dive deeper into each of these items.