Correct Answer :- B and D

Explanation:

Explanation from Amazon documents:

AWS Database Migration Service (AWS DMS) is a service that helps you migrate data from one data

source to another. AWS DMS supports full load and change data capture (CDC) modes, which enable

you to migrate data with minimal downtime. AWS DMS also supports parallel load, which allows you

to load data from multiple tables or partitions concurrently.

To improve database performance, reduce data migration time, and create multiple DMS tasks, the

database specialist should use the following combination of changes:

Use a smaller set of tables with each DMS task. Set the MaxFullLoadSubTasks parameter to a higher

value. This change will allow the database specialist to split the migration workload into smaller and

more manageable units, and increase the parallelism of the full load process. The

MaxFullLoadSubTasks parameter specifies the maximum number of tables that are loaded in parallel

for each DMS task. By setting this parameter to a higher value, the database specialist can increase

the throughput and performance of the full load process.

Use parallel load with different data boundaries for larger tables. This change will allow the database

specialist to divide the larger tables into smaller chunks based on a partition key or a range of values,

and load them in parallel using multiple DMS tasks. Parallel load can significantly reduce the

migration time and improve the performance of large tables.

Therefore, option B and D are the correct combination of changes to meet the requirements. Option

A is incorrect because increasing the value of the ParallelLoadThreads parameter in the DMS task

settings for the tables will not improve the performance or reduce the migration time significantly.

The ParallelLoadThreads parameter specifies the number of threads that are used to load data from a

single table or partition. By increasing this parameter, the database specialist may increase the CPU

utilization and network bandwidth consumption of the source and target databases, but not the

parallelism of the full load process. Option C is incorrect because using a smaller set of tables with

each DMS task and setting the MaxFullLoadSubTasks parameter to a lower value will decrease the

parallelism and performance of the full load process. Option E is incorrect because running the DMS

tasks on a larger instance class and increasing local storage on the instance will not address the root

cause of the performance issue, which is the lack of parallelism and partitioning of the large tables.