Exam AWS Certified Data Analytics - Specialty topic 1 question 125 discussion

Answer: C

Correct answer is C as only the last 4 months of data is required for 95%, the data can be stored in the Redshift cluster. The data covering the 12 months can be moved to S3 and queried using Redshift Spectrum. A mix of S3 and Redshift would provide the most cost-effective option. https://docs.aws.amazon.com/redshift/latest/dg/c-using-spectrum.html Option A is wrong as Redshift Spectrum cannot be used to query S3 Glacier storage data. Option B is wrong as using Redshift for all the data is not cost-effective. Option D is wrong as although Redshift federated queries would work, however, for 5% it would be cost-effective to query S3 directly instead of joining data from cluster and S3.

Point A cannot be answer because 5% user/query will be frequently used and Glacier need 90 to 180 days period. Point D : Federated query make sense when multiple data source because it required lots of authentication and authorization process , But here we have only S3 , so we can go with Point C

I don't get it: The question says: "System users will query the most recent 4 months of data 95% of the time while 5% of the queries will leverage data from the previous 12 months." 95% using data from the previous 4 month 5 % using data from the previous 12 month 0 % using data that is older than 12 month. So why not archive them?

C: I passed the test

Option C suggests storing the most recent 4 months of data in the Amazon Redshift cluster and using Amazon Redshift Spectrum to query data in the data lake, while ensuring that the S3 Standard storage class is in use with objects in the data lake. While this approach could work, it may not be the most cost-effective way to meet the requirements, because storing data in the Amazon Redshift cluster can be more expensive than storing data in S3. Additionally, by storing all data in the data lake, you may be able to use other data analysis services to query the data, which can be more cost-effective than using Amazon Redshift. Therefore, Option A, which leverages Amazon Redshift Spectrum to query the data in the data lake and uses S3 lifecycle management rules to move data from the previous 12 months to Amazon S3 Glacier, is likely a more cost-effective solution.

Selected Answer: C

C is the answer. If you have to join old data coming from S3 with new data coming from the Redshift cluster, you can do that. It is described here: https://catalog.us-east-1.prod.workshops.aws/workshops/e5548031-3004-49ad-89be-a13e8cd616f6/en-US/perform-analytics-on-your-data/join-and-query-data-with-redshift-spectrum

Not D. https://docs.aws.amazon.com/redshift/latest/dg/federated-limitations.html

Answer C should be correct

C. There is no mention of joining old and new data. Hence no need for federated queries.

Its C , textbook case for spectrum

I think D is correct. Using federated queries to combine redshift and Spectrum (from S3) will be more cost effective.

A - Data from past 12 months is not required so glacier is most cost effective, Redshift spectrum to be used for querying data 4 - 12 months older, and upto 4 months old data is to queried from Redshift DB

The question says "efficiently handle complicated queries", A. recommend S3 lifecycle, but I don´t think Glacier be efficient, even Expedited retrieval (1-5) minutes. I will go for C.

ali98 on point. Answer is C

Isn't D? The 5% of the user should be able to query both. moving the 4 months data to redshift makes sense and at the same time for 12 months data you need to query Redshift + Data lake data so in that federated queries can help to do that. In option C if they have provided Redshift Spectrum queries S3 as an external also then it make sense to choose C, But I do not read that way.