Exam AWS Certified Data Analytics - Specialty topic 1 question 22 discussion

I will go with A,C,E . Glue can do both pyspark and scala based ETL. Glue for Metadata and JDBC drivers to connect Athena from outside of AWS. Server less . so, Operational management is limited

Each word has meaning, So I will go with ABD, A metadata management that allows federation for access control, B- batch-based ETL using PySpark, D-JDBC connections from legacy clients. Not-C- because it mentioned only scala but questions mentioned scala operation is limited, E- you need JDBC to connect clinet not the Athena

Bing Option A is correct because AWS Glue Data Catalog provides a unified metadata repository across a variety of data sources and data formats, and it integrates with Amazon S3, Amazon RDS, Amazon Athena, Amazon Redshift, and others. Option B is correct because Amazon EMR with Apache Spark supports PySpark and Scala for batch-based ETL processing. Option E is correct because Amazon Athena supports SQL queries and can be integrated with JDBC drivers, allowing legacy clients to execute queries.

I will go with A. AWS Glue Data Catalog, B.Amazon EMR with Apache Spark, E. Amazon Athena aligns well with the company's requirements for a data lake architecture, offering a balance of performance, cost-efficiency, and ease of management. While C is also a viable option for ETL processes, it's more aligned with serverless ETL jobs and might not be as flexible for Scala as Amazon EMR with Apache Spark. D and F could provide JDBC connectivity and metadata management but are more operationally intensive and less integrated with S3 tiered storage strategies compared to using Athena with the Glue Data Catalog.

Why are we saying C ? C just says "Scala" ETL, even though Glue supports both pyspark and scala and AWS managed, the option specifically mentions "Scala based". Requirement is for both Scala and Pyspark that directly points to EMR. answer should be ABE.. about operational management, it says "limited", and EMR can qualify with it. using glue there is 'no' operational overhead.

ACE it

ACE: I passed the test

Correct answers are A, C & E Option A as Glue Data Catalog provides metadata management with the federation for access control. Option C as AWS Glue supports both serverless PySpark and Scala-based ETL with the least operational overhead. Option E as Athena can be used for querying S3 data. Athena can be connected using JDBC drivers from the external legacy clients. Options B, D & E is wrong as using EMR and RDS would increase the operational management and cost.

As operation management should be less, so all EMR related options are invalid, as EMR needs management of underlying EC2 instances

A. *Less* operational overhead compared to F (selected) B. High operational overhead, when compared to "C" AWS Glue based Scala C. *Less* operational overhead compared to "B" EMR PySpark (selected) D. Higher operational overhead when compared to "E" Athena. https://docs.aws.amazon.com/emr/latest/ReleaseGuide/HiveJDBCDriver.html E. *Less* operational overhead when compared to "D" EMR Hive JDBC (selected) F. Higher operational overhead when compared to "A" Glue metadata

I will go with A,C,E

D and F are wrong because the question never mentions Hive. E is not right, since Athena don't need JDBC to query S3. C is right because AWS Glue can be used for Scala-based ETL. A is right because Glue can connect on-premises DB through JDBC. https://aws.amazon.com/blogs/big-data/how-to-access-and-analyze-on-premises-data-stores-using-aws-glue/. B is right because Apache Spark can support PySpark.

My Answer is A, C & E

D, F are clearly wrong. D : JDBC connection from older clients is required, NOT from Athena. F : It is redundant to use RDS as a metastore. And there is no requirement for MySQL-compatible backed metastore. I'm torn between B and C. B is predominant in terms of cost constraints, but I am doubtful.

Answer: A,C,E EMR has operational overhead.

Answer: A,C,E EMR has operational overhead.

Ans - ACE Note: This is a free score question. Anything EMR comparing to serverless Glue / Athena is operational overhead. Also remember Glue can do PySpark and Scala, and Athena can do JDBC.