EC2 Instance Selection for SAP HANA — Key Factors

Quick Answer

Choosing an EC2 instance type for production SAP HANA comes down to matching the instance's characteristics to what an in-memory database actually needs to run reliably, and memory (RAM) size is the most fundamental of those factors because HANA keeps its working data set resident in memory rather than reading it from disk on every query - insufficient RAM means the database simply cannot hold the data it needs to operate, regardless of how fast the compute or storage is. Beyond raw memory, HANA also depends on consistent, low-latency I/O for persisting data changes and writing transaction logs, since even an in-memory database still needs durable storage for recovery and crash consistency. That's why EBS-optimized instances matter as another selection factor: dedicating network bandwidth specifically to EBS traffic prevents that storage I/O from competing with other network activity on the instance, which keeps log and data persistence performance predictable rather than variable under load. Together, these considerations point to a broader principle for sizing any in-memory database on AWS: you have to think about memory capacity for holding the working set, and separately about I/O consistency for the persistence layer, since a shortfall in either one can cause production problems even if the other is well provisioned. Whenever a question asks what to consider when picking an instance type for HANA, expect the answer set to span both memory sizing and I/O or network characteristics rather than compute power alone.