SAP has an
enormous amount of products, suites, acquisitions, and technologies. SAP Hana seems to integrate with many of them as an underlying platform. Its key selling feature is a drop-in replacement or complimentary side-by-side install with existing database solutions, with a migration. Just tossing in SAP Hana doesn't provide the benefits that actually optimizing the application for Hana would, and could actually be like throwing a Ferarri in the water like a boat anchor, instead of stripping out the engine and dropping it in the boat.
Since SAP Hana is an in-memory database, and it uses gobs of memory, you should probably have some knowledge of how memory is allocated. More info on
memory configuration in the Cookbook. If you're using Windows, Sysinternals RAMMAP is an invaluable resource for seeing what's going on with your memory. A
primer for memory in Linux is here. AWS has some
Cloudwatch scripts to monitor memory and disk usage.
If you miscalculate the RAM utilization, it sounds like out of memory errors could cause the platform to crash. In the case of Amazon AWS for Hana, the suggestion to fix here is
"did you turn it off and on?" or add on some expensive solid state hardware.
Another tip with memory usage is that row tables need to fit completely into memory, while column tables can cache only their indexes or indexes + part of the tables for performance. Row tables could cause problems if you decide to use Hana like a legacy database rather than a columnstore one. Hana is an
OLTAP database (Online Transactional & Analytical Processing).
Row vs. Column data source information here.
