A solutions architect is designing a data processing system that will use Amazon EC2 instances. Data that needs to be processed will wait in an Amazon Simple
Queue Service (Amazon SQS) queue. At least two data processing instances must run at all times.
Which combination of actions will meet these requirements MOST cost-effectively? (Choose two.)
The most cost-effective and reliable combination of actions is to purchase two Reserved Instances and to create a Spot Fleet with an appropriate scaling policy. Reserved Instances ensure that at least two instances are always running, satisfying the requirement of maintaining a minimum of two processing instances. The Spot Fleet, configured with a target scaling policy based on the acceptable backlog per instance, allows for cost-effective scaling by using lower-cost Spot Instances for additional capacity when needed. This allows the system to handle varying workloads efficiently while keeping costs down.
BD is right, 2 instances must run always so get RIs for that + spot fleet to compensate increased traffic
"At least two data processing instances must run at all times." So D does not make sense to me
Reserved two instances will always run anyway
A - OK. 2 instances must run at all times. Using spot can cause interruption. Can create custom metrics for Queue backlog per instance: https://docs.aws.amazon.com/autoscaling/ec2/userguide/as-using-sqs-queue.html#scale-sqs-queue-cli B - OK. RI is only applicable for on-demand instances - Reserved Instances are not physical instances, but rather a billing discount applied to the use of On-Demand Instances in your account. https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-reserved-instances.html Why not B & D? - Option D doesn’t meet requirement of 2 instances must run at all times due to spot interruption. Also RI will be useless since there is no on-demand instances.
It's a combination of actions. With option B, you are covering the basis of two instances running all the time, option D with the spot ASG it's for saving.
Both A and D will allow you to have 2 on-demand instances running full time (covered by the RIs bought in 'B') and scale with spot instances. The difference between the answers is the Spot fleet says it scales from the instance backlog while the ASG is scaling from the SQS queue. It's better to scale from the SQS queue, so I would choose B & D.
the question is asking "combination of actions", so BD.
AB because D is all spot which doesn't ensure there are always 2 instances are running. Also you need to actually have 2 on-demand instances running for the reserved instance discount in B to kick in - purchasing reserved instances alone doesn't result in anything running. https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/ec2-reserved-instances.html Reserved Instances are not physical instances, but rather a billing discount applied to the use of On-Demand Instances in your account.
I'll go with A and B. RI for 2 + Spots B and D make no sense. RI for nothing + Spots
BD is correct
for B and D https://docs.aws.amazon.com/autoscaling/ec2/userguide/ec2-auto-scaling-mixed-instances-groups.html
Great explanation caveman712. The combination of action is : A- You request 2 on demand instances + additional capacity with Spot B- You take RI for the 2 on demand instances.
BD https://docs.aws.amazon.com/autoscaling/ec2/userguide/ec2-auto-scaling-mixed-instances-groups.html#capacity-optimized-aws-cli
D is wrong. If you choose B,D, then you have two reserved instances and auto scaling group which manages spot instances. So why you need set minimum value to 2 for the spot instances taking account that you already have two reserved instances. it should set to 0. So A and B are right
Correct ab
A, B - this is the question about allocation strategy - more can be found here: https://docs.aws.amazon.com/autoscaling/ec2/userguide/ec2-auto-scaling-mixed-instances-groups.html
My choice is A&B
D is not correct because the instance must be available anytime.. so A B is the preferable
BD 100%
The key word is "combination of actions" and "cost effectively". RIs for the 2 you must have and spot for the rest.
2 always available rest spot. D vrs A: D uses SQS for scaling which makes sense. We end up with 2 RI + Spot as needed and available.
Its again wording and dicing for the answers... 2 RI plus rest spot and we probably want to utilize sqs queue length for scaling. Could be both: maybe AB, maybe BD. Dice it. One need to be psychic to know how to interpret the answers. Yes, order two RI to not used them makes no sense: exactly.
BD better than AB
BD is right, 2 instances must run always so get RIs for that + spot fleet to compensate increased traffic
AC gives the option of going for spot instances and the facility to launch EC2 incase of spot not being available. Purchasing reserved instances won't save cost in this case
my bad.. it must be BD
spot fleets allow you to select on demand capacity and set target tracking using CW metrics. "A" also specifies that target tracking is set to an acceptable backlog. https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/on-demand-in-spot.html https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/spot-fleet-target-tracking.html D is incorrect as you are setting the minimum value for spot to 2 rather than 0. It also states that scaling will be set to the size of the queue which I interpret as 1 instance to 1 message which seems a bit excessive and may be costly unless processing time per message is large.
A,B is optimal for costs, as D is fixing the autoscale to always have 2 other spots... so the reserved instances are for what?