EC2 High Memory u-series resilience

Question

A customer is interested in using u-6tb1 (bare metal) EC2 High Memory instances.  They are interested in the resilience of the instances. As they are not available for automatic EC2 recovery, the customers are asking:

1. What will happen to the data and the OS, if there is an outage of the instance? The hard drives will be moved to a new instance? Or they're all EBS volumes that can be attached to a new instance?
 2. What is our SLA on the replacing the instance? What is the expected downtime?

Igor · Accepted Answer

**Answer from SAP team:**

High memory instances use EBS in the same way as X1. If there is an outage of the dedicated host, a new one is automatically made available to the AWS Account. In addition, if dedicated host recovery is enabled, the u- instance will automatically get restarted on the replacement host in a similar manner to EC2 auto recovery.

We don’t provide an SLA for replacing the failed instance but we ensure there is spare capacity in each AZ. Typically we have seen it take around 15 – 30 mins for the recovery action. Once the recovery happens and the instance is started, the HANA DB needs to load back into memory.  With a u-6tb1 with EBS io1 volumes, this takes ~ 30 minutes

EC2 High Memory u-series resilience

Relevant content