I vividly remember my first data warehouse project. SAP BW 3.5 was just released and I was tasked with the objective of replicating billing documents from SAP R/3. It was a true learning experience and I was very proud to finally see data flowing into the infocubes. Fast forward a year and my initial haphazard approach was growing tough to support. Not to mention the fact that query performance was a common complaint. My manager at the time wisely took note and made the decision to send me to SAP BW310 – Data Warehousing. In this class, I was first introduced to the concept of the LSA and the layered approach to data warehouse architecture.
The Reporting Layer
SAP defines the reporting layer as a logical grouping of “objects that are used to perform queries for analysis.”
I can attest to the benefit and cost of an adequately performing reporting layer. On the benefit side, my users enjoyed acceptable query performance. However, the cost of that user satisfaction would require regular intervention on my part such as:
- Time-slice data into multiple infocubes or multiple dso’s
- Combine all data providers together under the union action of a multi-provider
- Build and manage multiple infocube aggregates
- Create secondary indexes for dso’s
Ultimately, the struggle to maintain marginal query performance cost increasingly more storage, supporting long-running overnight batch processes, daily effort to fix process chain failures, and additional complexity in future development cycles. The reporting layer had truly overgrown in complexity.
Reporting Layer Swap
I propose that anyone facing a similar scenario (but not ready to migrate an entire SAP BW system to SAP BW/4HANA) should consider transforming the reporting layer using SAP DWC. Since SAP DWC runs on SAP HANA and SAP DWC is a cloud service, one can quickly reap benefits like simplifying the data architecture and enhancing query performance.
To make the “swap” to SAP DWC, plan to replace data providers within the reporting layer with tables and views within a DWC space. Note: a DWC space is a container object (similar to an infoarea) that stores tables, views, and other objects.
A modified BW LSA architecture that incorporates SAP DWC spaces for the reporting layer might look like the following:
Before and After Example
See below for an example data flow illustrating how one may use SAP Data Warehouse Cloud as a reporting layer replacement and accelerator. Notice that the number of duplicate data copies reduce from five to two. In the BW LSA example, a permanent copy of the data exists in the harmonization layer, business transformation layer, and each object in the reporting layer. In the BW LSA + DWC example, a permanent copy of the data exists in the harmonization layer and once in DWC.
Looking back on those initial years of data warehouse development, I now appreciate the positive impact in-memory technology can have on a SAP BW system. The LSA concept works very well but eventually the underlying data storage technology becomes the bottleneck affecting all layers of the LSA. For a BW system running on a traditional database, SAP DWC may offer the right next step to restore performance and simplicity – all beginning with the reporting layer.
For more information on SAP Data Warehouse Cloud or to start today with a reporting layer that can drive innovation, check out the following links: