Broker / Operational Data Store (ODS) /
Warehouse Pattern

An Operational Data Store (ODS) is an informational database that typically stores detailed data about business events in order to support operational reporting needs and to act as a staging area for data warehouses and data marts used for data analysis. This pattern uses the Integration Broker Suite (IBS) to facilitate the aggregation of business events from a number of different applications and stage them within an ODS. This pattern is presented a logical view. The physical implementation of the pattern may actual be bus rather than a hub and spoke topology as shown.

Users can then use tools to query the ODS directly in order to get near real-time information to support operational processes and decision-making. The information can also be moved into a data warehouse in order to support higher levels of aggregation and more advanced analysis.

Data warehouse construction has typically been hampered by the need to extract information from a large variety of existing applications. But once a broker is in place - many transactions are flowing through it anyway - broker can be used to populate an ODS.

Please view the Broker / Operational Data Store (ODS) / Warehouse Pattern below:

• Lowers data warehouse implementation and maintenance costs.
• Provides a combined source of data in a standardized format for query and reporting use.
• Reduces the need for direct access to important transactional systems that may be overloaded.

• Some attributes of the source data may be lost when it is transformed and sent to the ODS.
• Historical data may be lost unless special steps are taken to preserve the view of data over time.

This pattern will be most useful in cases where there is a need to aggregate information in near real-time to support operational reporting and as part of an overall data warehousing infrastructure in support of an enterprise application.

In cases were near real-time currency of information is not required and where batch loads from the operational database into the data warehouse are feasible, the ODS may not be required.

As more transactions begin to flow through NIH's IBS, and multiple communities of users become interested in these events, an ODS capturing the events could become a useful addition to NIH's overall business intelligence capability.