This needs a blog, but for now:
- Upgrade your HANA revision. Rev.70 seems more frugal on memory
- If you run Rev.70 then migrate the statistics server, it will use less RAM
- Never use the row store, those tables always reside in memory
- Partition your tables, HANA only loads active partition tables
- Set your tables to load on demand, not at startup
- Use less data or buy a bigger AWS instance
- Code better models that materialise less data. This is often a problem when beginners write Calc Views or SQLScript
- Use Analytic Views and Union with Constant Values rather than joins
John