WWW Improve Your Business

Turbocharging Data Access with SAP-BI Accelerator

Naeem Hashmi, Chief Research Officer at Information Frameworks recently published a paper on the new SAP-BI Accelerator that caught my eye. Below is my summary of the points he makes. For a complete copy of his paper click here.

Building a data warehousing architecture requires following a process that begins with defining user requirements as clearly as possible. This too is the case with SAP-BI (formerly SAP-BW). The reason that user requirements are so crutial to define correctly, is that the design and development of InfoCubes is key to how data will be accessed and information presented to the end user.

With a well designed data model, performance has not been an issue for most smaller SAP clients. Any performance issues could have been managed by aggregating data and user data navigation views. However, in larger applications with complex data structures - with time based, multiple-hierarchies, snow-flaked dimensions - data access performance degrades rapidly.

"Now Generally Available BI Accelerator Complements SAP NetWeaver® and Enables Customers to Analyze Large Amounts of Critical Business Information up to 200-Times Faster than Alternative Tools."

SAP-BI Accelerator is an appliance (blade server) that sits alongside of the BI instance and that works with SAP-BI. Indexes and data are moved to the accelerator that is able to search indices and associated data at lightening speed.

With the SAP Netweaver technology, BI Accelerator serves as a cache at the SAP-BI application server and is at the same time a full InfoCube as well as an Aggregate. To the end user, this application does ad hoc data aggregation at high speed, on the fly, off the Cube based on user queries.

SAP-BI accelerator eliminates the need for aggregates in the data model. So SAP-BI accelerator will simplify data modeling, but will probably introduce (in the short term) some challenges in managing/ changing established processes and best practices. This does not do away with the need to based your design off user requirements, because you still need to build your Cubes, but it will certainly cut down on support costs due to ongoing development of views based on changing user requirements.

In the long term the intention of SAP is to make cubes optional ... but until then we will have to keep them in mind when developing data warehousing solutions in SAP.

1. Top 3 Myths of Business Intelligence
2. Data Rich, Information Poor

No comments: