AWS Certified Solutions Architect - Associate SAA-C03 Exam - Question 714

Question

A company runs a highly available web application on Amazon EC2 instances behind an Application Load Balancer. The company uses Amazon CloudWatch metrics.

As the traffic to the web application increases, some EC2 instances become overloaded with many outstanding requests. The CloudWatch metrics show that the number of requests processed and the time to receive the responses from some EC2 instances are both higher compared to other EC2 instances. The company does not want new requests to be forwarded to the EC2 instances that are already overloaded.

Which solution will meet these requirements?

Examice · Accepted Answer

To prevent overloading specific EC2 instances with too many outstanding requests, the least outstanding requests algorithm should be used. This algorithm routes new requests to the EC2 instance with the lowest number of in-progress requests, which helps balance the load more effectively and prevents any single instance from becoming a bottleneck. The metrics 'RequestCountPerTarget' and 'ActiveConnectionCount' are instrumental in determining the current load on each instance. 'RequestCountPerTarget' gives insight into the number of requests each target (EC2 instance) is handling, and 'ActiveConnectionCount' provides information on the number of active connections, helping to ensure that overloaded instances are bypassed in favor of those with lower loads.

h0ng97_spare_002 · Answer

Option B is correct because can use "RequestCountPerTarget" to identify the amount of requests for each EC2 instance. Then use "least outstanding requests algorithm" to route to targets with the lowest number of in progress requests.

Option D is wrong because "RequestCount" cannot identify the amount of requests for each EC2 instance. "RequestCount" is for the whole ALB.

Andy_09 · Answer

Option B would be the correct choice

Kezuko · Answer

https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-cloudwatch-metrics.html

To understand the types

osmk · Answer

The least outstanding requests routing algorithm routes requests to the targets with the lowest number of in progress requests > https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-target-groups.html

haci · Answer

The question is not asking for better performance in response time. It is just simply asking to distribute the number of requests among instances. So B seems more logical.

TruthWS · Answer

Option B

Moon239 · Answer

Why not D?

alawada · Answer

distribute the number of requests among instances

dkw2342 · Answer

IMO the correct answer is option D:

This is from an earlier version of the AWS documentation on ALB target groups - for some reason they removed this information in the current revision:

"Consider using least outstanding requests when the requests for your application vary in complexity or your targets vary in processing capability. Round robin is a good choice when the requests and targets are similar, or if you need to distribute requests equally among targets. You can compare the effect of round robin versus least outstanding requests using the following CloudWatch metrics: RequestCount, TargetConnectionErrorCount, and TargetResponseTime."
 https://web.archive.org/web/20200426172626/https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-target-groups.html#modify-routing-algorithm

Scheldon · Answer

AnswerB

osmk · Answer

D>>> The least outstanding requests routing algorithm routes requests to the targets with the lowest number of in progress requests > https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-target-groups.html

xBUGx · Answer

I think TargetResponseTime is the best indicator for telling is a server is overloaded or not

FlyingHawk · Answer

D will be correct if change ReqeustCount to RequestCountPerTarget.  RequestCount applies to the ALB rather than individual targets, making it unsuitable for identifying overloaded instances.
ActiveConnectionCount:

This metric indicates the current number of active TCP connections. It complements RequestCountPerTarget by helping identify instances that might be overwhelmed with connections, even if their request count seems moderate.

sk1974 · Answer

I will go with D since the question asks for the target response.

sk1974 · Answer

Actually ,I now vote for B and not D

chest_jd · Answer

D is correct. 
https://docs.aws.amazon.com/elasticloadbalancing/latest/application/load-balancer-cloudwatch-metrics.html
ActiveConnectionCount - The total number of concurrent TCP connections active from clients to the load balancer
RequestCountPerTarget - The average request count per target, in a target group
Both metrics do not show stage of every target.

RequestCount  - The number of requests processed over IPv4 and IPv6. This metric is only incremented for requests where the load balancer node was able to choose a target.
We can say how many requests go to targets.
TargetResponseTime  - The time elapsed, in seconds, after the request leaves the load balancer until the target starts to send the response headers. 
We have metric to get statistics for every target/instance.

In fact Item D is correct

AWS Certified Solutions Architect - Associate SAA-C03 Exam - Question 714

Discussion