I've been working on a project to convert a OLTP like SQL Server application to SAP HANA and I'm not getting the results that I expected. When using SAP HANA row tables with traditional indexes, I'm getting a 15% performance increase over SAP HANA column tables. I know that the use of row based tables is not sustainable because of the memory restrictions. I did run with the same indexes on the column tables. My suspicion is that I should probably remove all the indexes on my column tables and re-run the tests based on what I read in the ACM paper at http://www.sigmod09.org/images/sigmod1ktp-plattner.pdf.
What I'm really looking for is guidance on what to look for in an execution plan and what strategies can I use based on what I see in the plan. Is there any guidance out there on tuning column tables?
Regards,
Bill