Exam AWS Certified Solutions Architect - Associate SAA-C03 topic 1 question 581 discussion

By setting the Auto Scaling group's minimum capacity to four, the architect ensures that there are always at least two running instances. Deploying two On-Demand Instances in each of two Availability Zones ensures that the application is highly available and fault-tolerant. If one Availability Zone becomes unavailable, the application can still run in the other Availability Zone.

My rational is: Highly available = 2 AZ, and then 2 EC2 instances always running is 1 EC2 in each AZ. If an entire AZ fails, SacalinGroup deploy the minimun instances (2) on the running AZ

Option A: Set the Auto Scaling group's minimum capacity to two. Deploy one On-Demand Instance in one Availability Zone and one On-Demand Instance in a second Availability Zone. This configuration ensures that you have two instances running across two different AZs, which provides high availability. However, it does not take advantage of additional capacity to handle failures or spikes in demand. If either AZ becomes unavailable, you will have one running instance, but this does not meet the requirement of having at least two running instances at all times.

Set the Auto Scaling group's minimum capacity to four. Deploy two On-Demand Instances in one Availability Zone and two On-Demand Instances in a second Availability Zone. This configuration provides high availability with four instances distributed across two AZs. The minimum capacity of four ensures that even if one instance fails, there are still two instances in each AZ to handle the load. This option is highly available and fault-tolerant but may be more than required if only two instances are needed to be always running.

Answer is A: If one Availability Zone fails, the Auto Scaling group will automatically launch a new instance in a different, healthy Availability Zone to maintain the desired capacity of two instances. This is one of the key benefits of using Auto Scaling groups—they automatically maintain the desired number of instances across multiple Availability Zones, ensuring that your application is highly available and fault-tolerant. So even in the event of a failure in one Availability Zone, your application will continue to run on the required number of instances. This is why it’s recommended to distribute instances across multiple Availability Zones when designing architectures for high availability and fault tolerance.

so indeed ASG can set up a new EC2 instance in another AZ if there is one AZ failed with fault but it failed to meet the need of always having 2 instance running before the new instance replacement is done in the working AZ. so this is why we deploy 2 instances per AZ

If it would not mention the "stateful" application, and if it would only have to be "highly available" but NOT "fault-tolerant", A would be fine.

From <https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-best-practices.html : Spot Instances are not suitable for workloads that are inflexible, stateful, fault-intolerant, or tightly coupled between instance nodes. So C and D don't fit. From <https://docs.aws.amazon.com/whitepapers/latest/real-time-communication-on-aws/use-multiple-availability-zones.html : Within the constructs of AWS, customers are encouraged to run their workloads in more than one Availability Zone. This ensures that customer applications can withstand even a complete Availability Zone failure - a very rare event in itself. So a HA solution in this case implies a total of 4 instances, 2 per AZ.

The main requirement here is a 'highly available and fault-tolerant architecture for the application', this covered by option B. The application requires at least two EC2 instances to always be running, main word here being 'atleast' which means more than two is ok.

B - Need 2 in each AZ and you cant use spot instances as it could be recalled.

Stateful is keyword here. 2 is minimum required all time.

If a complete AZ fails, autoscale will lunch a second EC2 in the running AZ. If that short period of time is not always, which is not, then the answer is B, but I would take my chances and select A in the exam xD because the application is highly available and fault-tolerant.

° Minimum of 4 ensures at least 2 instances are always running in each AZ, meeting the HA requirement. ° On-Demand instances provide consistent performance and availability, unlike Spot. ° Spreading across 2 AZs adds fault tolerance, protecting from AZ failure.

While Spot Instances can be used to reduce costs, they might not provide the same level of availability and guaranteed uptime that On-Demand Instances offer. So I will go with B and not D.

Highly available - 2 AZ and then 2 EC2 instances always running. 2 in each AZ.

Highly available - 2 AZ and then 2 EC2 instances always running. 2 in each AZ..