Question

A company is deploying a new latency-sensitive SAP HANA in-memory database on AWS to support a real-time OLTP workload. The database requires very large amounts of RAM (over 6 TB) and consistent sub-millisecond storage response for its persistence and log volumes. The architecture team must design the compute and storage layer to meet SAP's certified performance requirements while keeping write latency as low and predictable as possible during peak transaction volumes. Which combination best satisfies these requirements?

Accepted Answer

Deploy on EC2 High Memory instances (u-*.metal / u-*tb1) certified for SAP HANA, and use provisioned IOPS SSD (io2 Block Express) EBS volumes for the data and log volumes.

Answer

Deploy on memory-optimized R6g Graviton instances and place the HANA data and log files on general purpose (gp3) EBS volumes with the maximum burst credits enabled.

Answer

Deploy on compute-optimized C6i instances behind an Auto Scaling group and store the data and log files on an Amazon EFS file system mounted across instances for elasticity.

Answer

Deploy on EC2 High Memory instances and store the HANA data and log volumes on Throughput Optimized HDD (st1) volumes to maximize sequential throughput at low cost.