Home › Forums › AWS › AWS Certified Solutions Architect Associate › Understanding ASG behaviour › Reply To: Understanding ASG behaviour
-
Hello, if this thread can be re-lived 🙂
I just stumbled on the same question. The explanation says that the option with 2-6 instances in 2 AZs “is incorrect. It is required to have 2 instances running all the time. If an AZ outage happened, ASG will launch a new instance on the unaffected AZ. This provisioning does not happen instantly, which means that for a certain period of time, there will only be 1 running instance left.”;
I don’t fully agree with such an explanation. In such logic, we can imagine that in times of peak load an AZ failure also can happen, leading to only 3 running instances left, which is also insufficient.
I believe that AWS advocates right-sizing and scaling dynamically, not pre-provisioning excess capacity as in the suggested answer. Manually overprovisioning does not solve this either. A 3-instance loss due to AZ failure would still result in insufficient capacity until ASG compensates.While I understand that the question might anticipate a solution with 0 downtime in case of the AZ failure, it would make more sense to provision a minimum of 3 instances in 3 AZs, which will be at least more cost-effective.