Exam AWS Certified Big Data - Specialty topic 1 question 44 discussion

Correct Answer is A The main asks are persistence and cost-efficiency. Let us calculate the storage and R/W numbers 2KB/File * 1 Million files/Hr * 24 Hrs * 30 Days = 1 Month of data 2/(1024 * 1024) * 1,000,000 * 24 * 30 = 1,373 GB of data/month 1,000,000 * 24 * 30 = 720 Million writes/month or ~278 writes/second. A) The cost of storing 1.37 TB of data in DynamoDB is $370. On-demand cost of writing ~278/sec into DynamoDB is ~600 a month with a total cost of ~$1000. It is persistence storage. B) The cost of storing 1.37 TB of data in S3 is ~$31. Put request cost of 720 Million objects ~$3600 a month with aa total cost of ~$3631. C) Storing 720 Million records with total size 1.3 TB in Elasticache memory will cost more than $10K/month along with snapshot cost for persistence. D) S3 IA costs are even higher than standard costs given that the minimum size requirement/object is 128KB. A is the most cost-efficient of all along with persistence SLA.

for N.virginia DynamoDB per GB 0.25/month + RCU S3 Standard per GB 0.023/month S3 Standard IA per GB /0.0125 but bill minimum 128KB for each object S3 Standard is cheaper

This article can be a clue. The answer could be A due to too many small files. https://www.reddit.com/r/aws/comments/5haamf/has_anyone_done_the_costs_math_on_s3_vs_dynamodb/

A is correct : Too many small files.. S3 is not a good choice. DynamoDB is the right option.

D. You can run a batch job e.g. every 30 minutes to gzip new files into one gz file and upload to S3 IA. All requirements met, only 48 API S3 PUT calls per day, and the file will for sure be bigger than 128kb.

B is the correct, for dynamo you also need to pay by request. Also you can call once per minute

Correct Answer is A

As the question is asking to store files on AWS, I will go with Option - B

my selection A

D would give you 30 days of storage per file, it doesn't say you have to keep the files, dynamoDB with TTL is a better option?

S3 Standard is most cost efficient. 2M Kb files per hour which leads to 48GB per day and 1152 GB per month For this S3 cost 28 USD DynamoDB cost 308 USD

I would agree with @pra276 and select A because too many small object put requests to S3 increases cost compared to DynamoDB.

I choose D The question is asking for "persistent storage layer", but DynamoDB is naturally a database. So I think S3 should be option. However, the question does not clearly mention how frequently the stored data need be accessed, so I will go with S3 IA which is less expensive than S3 standard.

D, Persistence and Cheap solution. The question didn't tell how frequent the files will be accessed. So D instead of B.

B correct

Answer is B as it is frequently accessed storage and cheaper

I'd look at the two key phrases in this question: a persistent storage layer AND most cost-efficient option. That leads me to D