DSS Application Examples
The DSS/BI examples start with a traditional star schema data warehouse deployment for a Silicon Valley high-tech company. The same data has also been deployed as an OLAP cube created by Analysis Services.
The last example, describes a hybrid distributed reporting system that, uses multiple SQL Server technologies such as data replication, database mirroring, and database snapshots to get the most out of a complex healthcare industry application environment.
DSS Example One
A Silicon Valley computer company implemented a traditional data warehouse using a star schema approach. A star schema provides multiple paths (dimensions) to the central data facts. As you can see in Figure 3.4, a decision support user can get to the Sales Units, Sales Price, and Sales Returns through Geographic, Time, and Product dimensions. This allows the user to ask questions such as "What were net sales for North America for a particular month for a specific computer product?" SQL Server Integration Services (SSIS) packages populate this data warehouse and conformed dimensions on a daily basis with deltas (new data changes only). The data warehouse is unavailable for about one hour each night as daily updates are rolled into this deployment.
Figure 3.4 Star schema data warehouse for global computer sales.
This SQL Server instance is isolated from the OLTP application where the data is sourced. There are about 500–600 data warehouse users of this data globally. This data warehouse is approaching 5TB in size.
DSS Example Two
The same Silicon Valley computer company also implemented some of the same data in a more complex Analysis Services OLAP cube for data mining purposes. The company had many things it did not know about its sales data and wanted to do complex trending and forecasting to better understand the demand for products worldwide. Figure 3.5 shows the OLAP cube built in Analysis Services for this complex business intelligence purpose. Several demand forecasting and product sales trending models were developed to allow this company to predict sales by each of its products for each geographic region.
Figure 3.5 Multidimensional OLAP cube in Analysis Services.
DSS Example Three
This last example features a multitechnology hybrid data reporting solution that provides real-time reporting along with point-in-time reporting for a major healthcare organization in the Pacific Northwest. This solution starts with real-time data replication from its online transactional systems where all hospital transactions are taking place. This includes patient events, medications administered, surgeries done, hospital charges, and so on. By distributing this data to a highly available two-node SQL Cluster, the hospital is able to realize all its real-time reporting requirements that center around providing all known information for a particular patient in the hospital at any time. Figure 3.6 shows this OLTP-to-SQL cluster real-time, continuous data replication and the real-time reporting enabled by this data distribution.
Figure 3.6 Hybrid SQL Server reporting configuration.
Another major reporting requirement for this health organization is not a real-time requirement, but rather a leisurely hourly snapshot, point-in-time reporting requirement. A much larger group of users must be served by this noncritical reporting need and cannot impact the real-time reporting environment in any way. To satisfy this point-in-time, noncritical reporting need, the health organization leveraged SQL Server database mirroring from the replicated SQL Server Health Provider DB. From the mirror, hourly database snapshots are created to satisfy all the point-in-time reporting needs of the organization. This configuration has been extremely stable since the SQL Server 2005 deployment.