As shown in Figure 2-18 on page 83, we also used 25 of this type of InfoCube (type B) for reporting, again having their aggregates filled and activated. This resulted in a combined total of 43 million*25 rows in the fact table and 21 million*25 rows across the aggregate tables, which yields a combined total of 1.60 billion rows. 2.6 SAP NetWeaver BI processes In order for the BW scalability benchmark to provide any meaningful and re-usable results, the specific processes defined for the benchmark needed to be based on an existing Benchmarking productive implementation. In this way, we could verify that the results obtained in the the core SAP benchmark matched the present experience. This would also give credibility to the NetWeaver BI benchmark results achieved when scaled from 20 TB to 60 TB in Phase 2 (which is not processes: the documented in this book). upload, the query and the rollup. Keep in mind, however, that the specific processes benchmarked here only represent a subset of the processes that a production system would be subject to. Nevertheless, we believe these processes either define the core processes, or processes that may be exposed to potential bottlenecks. The following processes were benchmarked: The upload process In this process, data is uploaded from ODS to target InfoCubes. On its way, it is subjected to expensive transformation rules implemented in the start routines. The query process In this process, the BW system is subject to users running queries.