Making the Case for Business Intelligence Solutions Powered by SAP HANA


I recently had the opportunity to present the afternoon keynote at the ASUG Georgia chapter meeting. The presentation was given on September 28th 2012. Attached is the presentation (as a PDF) for your review. The presentation is very high level and might not be a great source for a technical understanding of how SAP HANA works. On the other hand, if you are looking to gain a high level overview of SAP HANA and examples of its benefits this presentation will be appropriate. It would also recommend you scroll down and review the comments or make comments if you have any questions. Please note that your email address is not exposed to the public when leaving a comment.

Download the Presentation

 

View Video of my Interview with ASUG News

 

2 comments

  1. Jonathan,

    Thanks for excellent presentation. I loved it but have few questions for my understanding purposes.

    Slide 9/10/11: Could you share data volumes tested like how many records in BW, Cells in BEx and BOBJ ?

    Slide 12 : Could you compare the cost of SAN Memory vs HANA as you mentioned SAN is expensive ?

    Slide 12 : Can HANA play any role to improve performance with NLS (assume NLS for BW on HANA) as data is not InMemory ?

    Slide 13 : HANA also has overhead similar to Legacy DB Overhead & Change Log, why they were not included here ?

    Vamsi

  2. Slide 9/10/11: Could you share data volumes tested like how many records in BW, Cells in BEx and BOBJ ?

    I believe the CRM DSOs had over 12 Million rows, work order DSO had over 6 million. From the BOBJ side, most of the reports were limited to no more than 1000 rows (max). There were several reports included in the testing ranging from 10 columns to 50+ columns. Query Striping was enabled at both the query and report layers. We used BEx direct binding to BICS for all tests. Just as an FYI, if your reports contain 250,000 + total cells, don’t expect SAP HANA to solve any overall performance problems. BW on HANA will generate the query results much faster but once they are returned to Web Intelligence, processing and management of the data is given to the Web Intelligence processing server. In short, Web Intelligence in an analysis tool not a data extraction tool. Push down thoes operations to BW…

    Slide 12 : Could you compare the cost of SAN Memory vs HANA as you mentioned SAN is expensive ?

    First, SAN (Storage Area Network) is for persistent storage not memory. The client had invested well over 1 million dollars in several large 20-50 TB NetApp SANs that incorporated both magnetic 15K SAS drives and SSD for fast caching. They had several BW environment for regression and life-cycle testing purposes which required multiple devices. The large and fast SAN was implemented to help improve the performance of the underlying DB2 RDMS that was running BW. As you can see from the slides, hardware alone was not enough to make BW running on DB2 faster. The initial cost and maintenance of the SAN was more than the initial cost and maintenance of each SAP HANA multi-server cluster. Over time they would need to keep buying new SAN devices or upgrading the SAN. However, the cost to performance gain was starting to move in the negative. Performance was decreasing (as data volumes increased) while costs for the SAN increased.

    At another client site we also found that the software and maintenance cost of Oracle exceeded that of the software, hardware and maintenance cost of SAP HANA. Each client will be a unique case but the slides were intended to point out a few obvious cost saving areas to look for. When you combine that with the performance gains and user experience, BW on SAP HANA is a winner.

    Slide 12 : Can HANA play any role to improve performance with NLS (assume NLS for BW on HANA) as data is not InMemory ?

    NLS is not in-memory and SAP HANA has no direct relationship with it. It is an option in SAP BW NW 7.3 that allows you to move historical or infrequently accessed infoproviders from SAP HANA (or other RBMS running BW) to a special storage tier running on its own server. It incorporates columnar storage of data so you will find it to be faster (when configured correctly) than BW on a legacy RDMS. Most client are looking at using NLS to further reduce the TCO of SAP HANA by moving small volume Infoproviders or historical data into NLS as opposed to SAP HANA’s in-memory engine. for more information check out Dolphin’s BW arching solution

    Slide 13 : HANA also has overhead similar to Legacy DB Overhead & Change Log, why they were not included here ?

    There is a difference in BW powered by SAP HANA. Check out Dr. Bjarne Berg of Comerit and Penny Silvia of IBM interview with Scott Wallask, SAPexperts editor for more information.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s