Non relational systems commonly do not enforce referential integrity automatically which means developers must manage related data consistency themselves is the correct choice.

Non relational systems commonly do not enforce referential integrity automatically which means developers must manage related data consistency themselves captures the key disadvantage that many document and key value stores do not provide built in foreign key constraints or automatic referential checks. That means the application must implement logic to prevent orphaned records and to keep related documents consistent which increases development and testing effort and raises the risk of subtle bugs.

Relational databases are often a better choice for workloads that demand very high transactional throughput is incorrect because it states a generalization about throughput. Throughput depends on the specific database design and scaling approach and many non relational systems are optimized for very high write and read throughput at large scale, so this statement is not a reliable disadvantage of non relational stores.

Document and key value stores generally support flexible schemas which makes altering the data model simpler is incorrect because it describes an advantage rather than a disadvantage. Flexible schemas are a common benefit of document and key value stores since they avoid costly schema migrations and allow records to evolve independently.

Some NoSQL solutions do not provide strong multirow transactional guarantees which can complicate atomic updates across multiple records is misleading as the best answer here. Historically that limitation existed in some NoSQL systems, but many modern document databases and distributed databases now provide multi document transactions or other consistency models, and the question focuses on the lack of automatic referential integrity which is a more fundamental disadvantage for relational features.

Full AWS Practitioner Certification Question