Q22 — AWS SAP-C02 Ch.3

Question 22 of 75 | ← Chapter 3

Q247. A company ingests and processes streaming market data. The data rate is constant. A nightly process that calculates aggregate statistics is run, and each execution takes about 4 hours to complete. The statistical analysis is not mission critical to the business, and previous data points are picked up on the next execution if a particular run fails. The current architecture uses a pool of Amazon EC2 Reserved Instances with 1-year reservations running full time to ingest and store the streaming data in attached Amazon EBS volumes. On- Demand EC2 instances are launched each night to perform the nightly processing, accessing the stored data from NFS shares on the ingestion servers, and terminating the nightly processing servers when complete. The Reserved Instance reservations are expiring, and the company needs to determine whether to purchase new reservations or implement a new design. Which is the most cost-effective design?

Correct Answer: B. Update the ingestion process to use Amazon Kinesis Data Firehouse to save data to Amazon S3. Use AWS Batch to perform nightly processing with a Spot market bid of 50% of the On-Demand price.

Explanation

Using Amazon Kinesis Data Firehose to save data to Amazon S3 simplifies the ingestion process and can reduce costs by eliminating the need for attached EBS volumes on EC2 instances. AWS Batch allows the company to run batch processing jobs on Spot instances, which can significantly reduce the cost of running these jobs compared to On-Demand instances. By setting a maximum bid of 50% of the On-Demand price, the company can have confidence that they are still getting a good deal while also ensuring that their jobs will complete in a timely manner. Option A is less optimal than Option B because it uses a fleet of On-Demand EC2 instances instead of Spot instances. This would likely be more expensive than using Spot instances with AWS Batch. Option C is less optimal than Option B because it uses Reserved Instances with longer leases (3 years) instead of Spot instances. Spot instances provide greater flexibility and cost savings compared to Reserved Instances. Option D is less optimal than Option B because it uses Amazon Redshift, which is a more expensive service than Amazon S3. Additionally, running an AWS Lambda function nightly may add unnecessary costs, as AWS Batch with Spot instances can handle the same workload at a lower cost. A.More expensive than B B.As it is not mission critical and can pick up from previous data point, Spot instance makes sense C.If we still use EBS, each instance will have its own EBS and data is hard to aggregate. EC2 is expensive as well D.Lambda has process limit of 15 mins