I wrote about SAP’s BI Accelerator quite a bit in my white paper on memory-centric data management, but otherwise I seem not to have posted much about it here. In essence, it’s a product that’s all RAM-based, and generally geared for multi-hundred-gigabyte data marts. The basic design is a compression-heavy column-based architecture, evolved from SAP’s text-indexing technology TREX. Like data warehouse appliances, it eschews indexing, relying instead on blazingly fast table scans.
I asked Lothar Schubert of SAP how BIA was doing in the market in its early going. This was his response:
BI accelerator today has been proven to work, and more importantly to live up to its promises. Several customers already went live with BI accelerator, and even more customers completed a POC already. The latter include names such as BP, Novartis, Brown-Foreman, and Coca Cola. The benefits customers cite are: (1) consistently fast performance, without dedicated tuning efforts, (2) new level of flexibility, (3) easy to deploy. Separately, in a lab test Winter Corporation confirmed the performance results, when they tested a 1.2 billion row retail scenario.