It can't exceed 900 RCUs, so "just below 900 RCUs" is "at capacity". Change to auto-scaling. If you were at 896-899, wouldn't you say "woah, that's too close!" or would you say "Must be lambda time-outs?" Best case, you'd still address dynamoDB even if it WAS lambda time-outs.

Amazon CloudWatch metrics show that the peak usage of the DynamoDB table is just below 900 RCUs. => There is NO Issue w.r.t DynamoDB So, A,B and D filtered out. ANSWER is B

A) chat gpt is right after verifying the message

Correct A.

Its about what is most "cost effective". So question asks what has most impact on costs? I would argue its A because while peak is above 900 its way below most of the time ( RCU overprovisioned). The may give cost savings while also solving the issue.

ChatGPT said: The most cost-effective solution to resolve the issue is option A: Configure the DynamoDB table's read capacity to use auto scaling with default parameters. By enabling auto scaling, the DynamoDB table can automatically adjust its read capacity based on the actual usage, which helps ensure that the table has sufficient read capacity to handle the traffic during peak hours without incurring unnecessary costs during off-peak hours. The other options may alleviate the issue, but they come at the cost of increased provisioned capacity, higher Lambda function timeouts, and additional resources to implement the data replication. By using auto scaling, the company can cost-effectively ensure that their system is able to handle the traffic during peak hours.

I'll go with A "Amazon DynamoDB auto scaling uses the AWS Application Auto Scaling service to dynamically adjust provisioned throughput capacity on your behalf, in response to actual traffic patterns. This enables a table or a global secondary index to increase its provisioned read and write capacity to handle sudden increases in traffic, without throttling." https://docs.aws.amazon.com/amazondynamodb/latest/developerguide/AutoScaling.html

"B" is the one.

RCU is not a problem

The answer is definitely not A. The key words are "with default parameters" Default auto-scaling read parameters have a maximum of 10. If the table is already using 900 RCUs, this would severely lower the resources. 900 -> 10

Customers are reporting timeout errors and slow performance.. Increasing timemout will caused it even slower. Thus going for A. At least it provide some read performance enhancement.

I think it's B.