Amazon AWS Certified Solutions Architect Associate SAA-C03 Database Models Replicas Connections and Caching Practice Test

Practice Exams:

View All

Amazon AWS Certified Solutions Architect Associate SAA-C03 Database Models Replicas Connections and Caching Practice Test

Topic 13 covers database models replicas connections and caching for the AWS Certified Solutions Architect – Associate certification. These original practice questions apply the verified SAA-C03 objectives to practical decisions and troubleshooting. Select one answer unless a fixed number is requested. For broader preparation, visit the AWS Certified Solutions Architect Associate SAA-C03 Exam Dumps page. Each option includes an explanation of the relevant behavior and scenario constraints.

Question 1

An order system requires multi-row transactions, foreign keys, and complex relational joins. Which data model is the best fit?

Use a graph database such as Amazon Neptune.
Use a relational database such as Amazon RDS or Aurora.
Use DynamoDB with a well-designed partition key.
Use a managed document-oriented database when the application needs flexible JSON-like document structures.
Use Amazon DynamoDB for the key-value workload.

Correct Answer: B

Correct Answer

Answer B is correct because Relational engines support SQL, constraints, joins, and transactional integrity across related tables. This directly meets the decisive requirement: relational database.

Incorrect Answers

Answer A is incorrect because Graph databases are optimized for traversing relationships among connected entities. It may be useful elsewhere, but it does not meet the decisive requirement here: relational database.

Answer C is incorrect because DynamoDB scales horizontally for high request rates when access patterns and partition keys distribute traffic effectively. It may be useful elsewhere, but it does not meet the decisive requirement here: relational database.

Answer D is incorrect because Document databases are designed around document retrieval and flexible nested attributes rather than normalized joins. It may be useful elsewhere, but it does not meet the decisive requirement here: relational database.

Answer E is incorrect because DynamoDB is a managed key-value and document database designed for low-latency access at large scale. It may be useful elsewhere, but it does not meet the decisive requirement here: relational database.

Question 2

A shopping-cart service performs simple key-based lookups at very high scale and does not need joins. Which database type is appropriate?

Use DynamoDB with a well-designed partition key.
Use a relational database such as Amazon RDS or Aurora.
Use a managed document-oriented database when the application needs flexible JSON-like document structures.
Use a graph database such as Amazon Neptune.
Use Amazon DynamoDB for the key-value workload.

Correct Answer: E

Correct Answer

Answer E is correct because DynamoDB is a managed key-value and document database designed for low-latency access at large scale. This directly meets the decisive requirement: DynamoDB.

Incorrect Answers

Answer A is incorrect because DynamoDB scales horizontally for high request rates when access patterns and partition keys distribute traffic effectively. It may be useful elsewhere, but it does not meet the decisive requirement here: DynamoDB.

Answer B is incorrect because Relational engines support SQL, constraints, joins, and transactional integrity across related tables. It may be useful elsewhere, but it does not meet the decisive requirement here: DynamoDB.

Answer C is incorrect because Document databases are designed around document retrieval and flexible nested attributes rather than normalized joins. It may be useful elsewhere, but it does not meet the decisive requirement here: DynamoDB.

Answer D is incorrect because Graph databases are optimized for traversing relationships among connected entities. It may be useful elsewhere, but it does not meet the decisive requirement here: DynamoDB.

Question 3

A content system stores JSON documents with flexible fields and retrieves whole documents by key. Which model best matches?

Use a managed document-oriented database when the application needs flexible JSON-like document structures.
Use a relational database such as Amazon RDS or Aurora.
Use DynamoDB with a well-designed partition key.
Use Amazon DynamoDB for the key-value workload.
Use a graph database such as Amazon Neptune.

Correct Answer: A

Correct Answer

Answer A is correct because Document databases are designed around document retrieval and flexible nested attributes rather than normalized joins. This directly meets the decisive requirement: document database.

Incorrect Answers

Answer D is incorrect because DynamoDB is a managed key-value and document database designed for low-latency access at large scale. It may be useful elsewhere, but it does not meet the decisive requirement here: document database.

Answer E is incorrect because Graph databases are optimized for traversing relationships among connected entities. It may be useful elsewhere, but it does not meet the decisive requirement here: document database.

Question 4

A fraud engine repeatedly traverses relationships among accounts, devices, and transactions. Which database model best matches the access pattern?

Use a graph database such as Amazon Neptune.
Use a managed document-oriented database when the application needs flexible JSON-like document structures.
Use a relational database such as Amazon RDS or Aurora.
Use DynamoDB with a well-designed partition key.
Use Amazon DynamoDB for the key-value workload.

Correct Answer: A

Correct Answer

Answer A is correct because Graph databases are optimized for traversing relationships among connected entities. This directly meets the decisive requirement: graph database.

Incorrect Answers

Answer B is incorrect because Document databases are designed around document retrieval and flexible nested attributes rather than normalized joins. It may be useful elsewhere, but it does not meet the decisive requirement here: graph database.

Answer C is incorrect because Relational engines support SQL, constraints, joins, and transactional integrity across related tables. It may be useful elsewhere, but it does not meet the decisive requirement here: graph database.

Answer D is incorrect because DynamoDB scales horizontally for high request rates when access patterns and partition keys distribute traffic effectively. It may be useful elsewhere, but it does not meet the decisive requirement here: graph database.

Question 5

A telemetry system needs massive horizontally scalable writes with predictable partition-key access. Which AWS database is a common fit?

Use a managed document-oriented database when the application needs flexible JSON-like document structures.
Use a relational database such as Amazon RDS or Aurora.
Use DynamoDB with a well-designed partition key.
Use Amazon DynamoDB for the key-value workload.
Use a graph database such as Amazon Neptune.

Correct Answer: C

Correct Answer

Answer C is correct because DynamoDB scales horizontally for high request rates when access patterns and partition keys distribute traffic effectively. This directly meets the decisive requirement: DynamoDB.

Incorrect Answers

Answer A is incorrect because Document databases are designed around document retrieval and flexible nested attributes rather than normalized joins. It may be useful elsewhere, but it does not meet the decisive requirement here: DynamoDB.

Question 6

An application depends on a PostgreSQL extension that Aurora does not support but Amazon RDS for PostgreSQL supports. Which engine choice should drive the migration?

Use Amazon Aurora with the required MySQL/PostgreSQL compatibility.
Use Amazon RDS for Microsoft SQL Server.
Use Amazon RDS or Aurora rather than self-managed database hosts.
Add read replicas or Aurora Replicas and direct eligible reads to them.
Choose the managed relational engine that supports the required database extensions and compatibility, such as RDS for PostgreSQL in this scenario.

Correct Answer: E

Correct Answer

Answer E is correct because Engine compatibility is a hard requirement when the application depends on features that are not available in another compatible-looking service. This directly meets the decisive requirement: RDS PostgreSQL if it supports the extension.

Incorrect Answers

Answer A is incorrect because Aurora provides a managed relational engine with distributed storage and supports scaling reads with Aurora Replicas. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS PostgreSQL if it supports the extension.

Answer B is incorrect because RDS for SQL Server provides the required SQL Server engine in a managed database service. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS PostgreSQL if it supports the extension.

Answer C is incorrect because Managed relational services automate many infrastructure operations such as backups, patching, monitoring, and high-availability configuration. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS PostgreSQL if it supports the extension.

Answer D is incorrect because Read replicas offload read queries from the writer while preserving the writer for transactional changes. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS PostgreSQL if it supports the extension.

Question 7

A MySQL-compatible workload needs a managed relational service with distributed storage and multiple low-lag readers. Which engine family is a strong fit?

Use Amazon Aurora with the required MySQL/PostgreSQL compatibility.
Add read replicas or Aurora Replicas and direct eligible reads to them.
Use Amazon RDS or Aurora rather than self-managed database hosts.
Use Amazon RDS for Microsoft SQL Server.
Choose the managed relational engine that supports the required database extensions and compatibility, such as RDS for PostgreSQL in this scenario.

Correct Answer: A

Correct Answer

Answer A is correct because Aurora provides a managed relational engine with distributed storage and supports scaling reads with Aurora Replicas. This directly meets the decisive requirement: Aurora.

Incorrect Answers

Answer B is incorrect because Read replicas offload read queries from the writer while preserving the writer for transactional changes. It may be useful elsewhere, but it does not meet the decisive requirement here: Aurora.

Answer D is incorrect because RDS for SQL Server provides the required SQL Server engine in a managed database service. It may be useful elsewhere, but it does not meet the decisive requirement here: Aurora.

Answer E is incorrect because Engine compatibility is a hard requirement when the application depends on features that are not available in another compatible-looking service. It may be useful elsewhere, but it does not meet the decisive requirement here: Aurora.

Question 8

A vendor application certifies only Microsoft SQL Server and requires its engine-specific features. Which managed AWS database should be selected?

Use Amazon Aurora with the required MySQL/PostgreSQL compatibility.
Use Amazon RDS for Microsoft SQL Server.
Choose the managed relational engine that supports the required database extensions and compatibility, such as RDS for PostgreSQL in this scenario.
Add read replicas or Aurora Replicas and direct eligible reads to them.
Use Amazon RDS or Aurora rather than self-managed database hosts.

Correct Answer: B

Correct Answer

Answer B is correct because RDS for SQL Server provides the required SQL Server engine in a managed database service. This directly meets the decisive requirement: RDS for SQL Server.

Incorrect Answers

Answer C is incorrect because Engine compatibility is a hard requirement when the application depends on features that are not available in another compatible-looking service. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS for SQL Server.

Answer E is incorrect because Managed relational services automate many infrastructure operations such as backups, patching, monitoring, and high-availability configuration. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS for SQL Server.

Question 9

A relational workload is read-heavy and the writer is not saturated. Which design increases read capacity without moving writes?

Use Amazon Aurora with the required MySQL/PostgreSQL compatibility.
Use Amazon RDS or Aurora rather than self-managed database hosts.
Add read replicas or Aurora Replicas and direct eligible reads to them.
Use Amazon RDS for Microsoft SQL Server.
Choose the managed relational engine that supports the required database extensions and compatibility, such as RDS for PostgreSQL in this scenario.

Correct Answer: C

Correct Answer

Answer C is correct because Read replicas offload read queries from the writer while preserving the writer for transactional changes. This directly meets the decisive requirement: read replicas.

Incorrect Answers

Answer B is incorrect because Managed relational services automate many infrastructure operations such as backups, patching, monitoring, and high-availability configuration. It may be useful elsewhere, but it does not meet the decisive requirement here: read replicas.

Question 10

A small team wants automated backups, patching, and managed high availability instead of administering database hosts. Which service category is appropriate?

Use Amazon Aurora with the required MySQL/PostgreSQL compatibility.
Choose the managed relational engine that supports the required database extensions and compatibility, such as RDS for PostgreSQL in this scenario.
Use Amazon RDS or Aurora rather than self-managed database hosts.
Add read replicas or Aurora Replicas and direct eligible reads to them.
Use Amazon RDS for Microsoft SQL Server.

Correct Answer: C

Correct Answer

Answer C is correct because Managed relational services automate many infrastructure operations such as backups, patching, monitoring, and high-availability configuration. This directly meets the decisive requirement: managed RDS/Aurora.

Incorrect Answers

Answer B is incorrect because Engine compatibility is a hard requirement when the application depends on features that are not available in another compatible-looking service. It may be useful elsewhere, but it does not meet the decisive requirement here: managed RDS/Aurora.

Answer E is incorrect because RDS for SQL Server provides the required SQL Server engine in a managed database service. It may be useful elsewhere, but it does not meet the decisive requirement here: managed RDS/Aurora.

Question 11

A DynamoDB table uses customerId as the partition key, but one customer generates half of all writes and causes throttling. What design issue must be corrected?

Create an appropriate global secondary index for the additional access pattern.
Use DynamoDB on-demand capacity mode.
Use eventually consistent reads when the application can tolerate them.
Use customerId as the partition key and order timestamp or another ordered value as the sort key.
Redesign the partition key or add write sharding so traffic is distributed across more partition values.

Correct Answer: E

Correct Answer

Answer E is correct because A skewed partition key can create a hot partition even when total table capacity is available elsewhere. This directly meets the decisive requirement: partition-key distribution.

Incorrect Answers

Answer A is incorrect because A secondary index provides an alternate key structure and avoids full table scans for a known query pattern. It may be useful elsewhere, but it does not meet the decisive requirement here: partition-key distribution.

Answer B is incorrect because On-demand mode automatically accommodates changing request traffic without requiring the customer to set read/write capacity units. It may be useful elsewhere, but it does not meet the decisive requirement here: partition-key distribution.

Answer C is incorrect because Eventually consistent reads can reduce read-capacity consumption compared with strongly consistent reads for the same item size. It may be useful elsewhere, but it does not meet the decisive requirement here: partition-key distribution.

Answer D is incorrect because A composite key groups related items by partition key and orders them by the sort key for range queries. It may be useful elsewhere, but it does not meet the decisive requirement here: partition-key distribution.

Question 12

An application needs all orders for one customer sorted by order timestamp. Which key design is appropriate?

Create an appropriate global secondary index for the additional access pattern.
Use eventually consistent reads when the application can tolerate them.
Use customerId as the partition key and order timestamp or another ordered value as the sort key.
Use DynamoDB on-demand capacity mode.
Redesign the partition key or add write sharding so traffic is distributed across more partition values.

Correct Answer: C

Correct Answer

Answer C is correct because A composite key groups related items by partition key and orders them by the sort key for range queries. This directly meets the decisive requirement: customer partition key plus timestamp sort key.

Incorrect Answers

Answer B is incorrect because Eventually consistent reads can reduce read-capacity consumption compared with strongly consistent reads for the same item size. It may be useful elsewhere, but it does not meet the decisive requirement here: customer partition key plus timestamp sort key.

Answer D is incorrect because On-demand mode automatically accommodates changing request traffic without requiring the customer to set read/write capacity units. It may be useful elsewhere, but it does not meet the decisive requirement here: customer partition key plus timestamp sort key.

Answer E is incorrect because A skewed partition key can create a hot partition even when total table capacity is available elsewhere. It may be useful elsewhere, but it does not meet the decisive requirement here: customer partition key plus timestamp sort key.

Question 13

A table is keyed by orderId, but a frequent query needs all open orders by account without scanning. Which DynamoDB feature should be added?

Create an appropriate global secondary index for the additional access pattern.
Redesign the partition key or add write sharding so traffic is distributed across more partition values.
Use DynamoDB on-demand capacity mode.
Use customerId as the partition key and order timestamp or another ordered value as the sort key.
Use eventually consistent reads when the application can tolerate them.

Correct Answer: A

Correct Answer

Answer A is correct because A secondary index provides an alternate key structure and avoids full table scans for a known query pattern. This directly meets the decisive requirement: secondary index.

Incorrect Answers

Answer B is incorrect because A skewed partition key can create a hot partition even when total table capacity is available elsewhere. It may be useful elsewhere, but it does not meet the decisive requirement here: secondary index.

Answer C is incorrect because On-demand mode automatically accommodates changing request traffic without requiring the customer to set read/write capacity units. It may be useful elsewhere, but it does not meet the decisive requirement here: secondary index.

Answer E is incorrect because Eventually consistent reads can reduce read-capacity consumption compared with strongly consistent reads for the same item size. It may be useful elsewhere, but it does not meet the decisive requirement here: secondary index.

Question 14

Traffic is unpredictable and ranges from almost zero to sudden large bursts. Which DynamoDB capacity mode reduces manual capacity planning?

Create an appropriate global secondary index for the additional access pattern.
Redesign the partition key or add write sharding so traffic is distributed across more partition values.
Use customerId as the partition key and order timestamp or another ordered value as the sort key.
Use eventually consistent reads when the application can tolerate them.
Use DynamoDB on-demand capacity mode.

Correct Answer: E

Correct Answer

Answer E is correct because On-demand mode automatically accommodates changing request traffic without requiring the customer to set read/write capacity units. This directly meets the decisive requirement: on-demand capacity mode.

Incorrect Answers

Answer C is incorrect because A composite key groups related items by partition key and orders them by the sort key for range queries. It may be useful elsewhere, but it does not meet the decisive requirement here: on-demand capacity mode.

Answer D is incorrect because Eventually consistent reads can reduce read-capacity consumption compared with strongly consistent reads for the same item size. It may be useful elsewhere, but it does not meet the decisive requirement here: on-demand capacity mode.

Question 15

A workload performs strongly consistent reads but wants twice the read throughput from the same capacity. Which property of eventually consistent reads is relevant?

Use customerId as the partition key and order timestamp or another ordered value as the sort key.
Create an appropriate global secondary index for the additional access pattern.
Use eventually consistent reads when the application can tolerate them.
Use DynamoDB on-demand capacity mode.
Redesign the partition key or add write sharding so traffic is distributed across more partition values.

Correct Answer: C

Correct Answer

Answer C is correct because Eventually consistent reads can reduce read-capacity consumption compared with strongly consistent reads for the same item size. This directly meets the decisive requirement: eventually consistent reads consume less read capacity per item.

Incorrect Answers

Answer A is incorrect because A composite key groups related items by partition key and orders them by the sort key for range queries. It may be useful elsewhere, but it does not meet the decisive requirement here: eventually consistent reads consume less read capacity per item.

Answer B is incorrect because A secondary index provides an alternate key structure and avoids full table scans for a known query pattern. It may be useful elsewhere, but it does not meet the decisive requirement here: eventually consistent reads consume less read capacity per item.

Question 16

An Aurora cluster has several replicas and the application should spread read-only queries across them automatically. Which endpoint should it use?

Use an Aurora Global Database or supported cross-Region read replicas for regional read locality.
Read from the authoritative writer or another data path that provides the required current/strong consistency.
Use the Aurora reader endpoint for eligible read-only connections.
Copy analytical data to a separate warehouse or analytical replica.
Run the reporting workload against a read replica.

Correct Answer: C

Correct Answer

Answer C is correct because The reader endpoint distributes connections among available Aurora Replicas rather than sending them to the writer. This directly meets the decisive requirement: reader endpoint.

Incorrect Answers

Answer A is incorrect because Cross-Region replicas can place read capacity closer to users while preserving a designated writer Region. It may be useful elsewhere, but it does not meet the decisive requirement here: reader endpoint.

Answer B is incorrect because A replica with asynchronous lag is not appropriate when the decision requires the latest committed state. It may be useful elsewhere, but it does not meet the decisive requirement here: reader endpoint.

Answer D is incorrect because Separating heavy analytical scans prevents them from competing directly with the transactional writer workload. It may be useful elsewhere, but it does not meet the decisive requirement here: reader endpoint.

Answer E is incorrect because A read replica offloads eligible queries when some replication lag is acceptable. It may be useful elsewhere, but it does not meet the decisive requirement here: reader endpoint.

Question 17

A reporting query can tolerate seconds of replica lag and is overloading the primary database. Which design should offload it?

Copy analytical data to a separate warehouse or analytical replica.
Run the reporting workload against a read replica.
Use the Aurora reader endpoint for eligible read-only connections.
Read from the authoritative writer or another data path that provides the required current/strong consistency.
Use an Aurora Global Database or supported cross-Region read replicas for regional read locality.

Correct Answer: B

Correct Answer

Answer B is correct because A read replica offloads eligible queries when some replication lag is acceptable. This directly meets the decisive requirement: read replica.

Incorrect Answers

Answer A is incorrect because Separating heavy analytical scans prevents them from competing directly with the transactional writer workload. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Answer C is incorrect because The reader endpoint distributes connections among available Aurora Replicas rather than sending them to the writer. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Answer D is incorrect because A replica with asynchronous lag is not appropriate when the decision requires the latest committed state. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Answer E is incorrect because Cross-Region replicas can place read capacity closer to users while preserving a designated writer Region. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Question 18

Global users need low-latency reads near their Regions while one Region remains the writer. Which architecture is appropriate for Aurora?

Use an Aurora Global Database or supported cross-Region read replicas for regional read locality.
Read from the authoritative writer or another data path that provides the required current/strong consistency.
Use the Aurora reader endpoint for eligible read-only connections.
Run the reporting workload against a read replica.
Copy analytical data to a separate warehouse or analytical replica.

Correct Answer: A

Correct Answer

Answer A is correct because Cross-Region replicas can place read capacity closer to users while preserving a designated writer Region. This directly meets the decisive requirement: Aurora Global Database or cross-Region replicas.

Incorrect Answers

Answer D is incorrect because A read replica offloads eligible queries when some replication lag is acceptable. It may be useful elsewhere, but it does not meet the decisive requirement here: Aurora Global Database or cross-Region replicas.

Answer E is incorrect because Separating heavy analytical scans prevents them from competing directly with the transactional writer workload. It may be useful elsewhere, but it does not meet the decisive requirement here: Aurora Global Database or cross-Region replicas.

Question 19

A nightly analytical workload runs large scans that interfere with transactions. Which design best isolates the analytical workload?

Run the reporting workload against a read replica.
Use the Aurora reader endpoint for eligible read-only connections.
Copy analytical data to a separate warehouse or analytical replica.
Read from the authoritative writer or another data path that provides the required current/strong consistency.
Use an Aurora Global Database or supported cross-Region read replicas for regional read locality.

Correct Answer: C

Correct Answer

Answer C is correct because Separating heavy analytical scans prevents them from competing directly with the transactional writer workload. This directly meets the decisive requirement: separate analytical copy/warehouse.

Incorrect Answers

Answer A is incorrect because A read replica offloads eligible queries when some replication lag is acceptable. It may be useful elsewhere, but it does not meet the decisive requirement here: separate analytical copy/warehouse.

Answer B is incorrect because The reader endpoint distributes connections among available Aurora Replicas rather than sending them to the writer. It may be useful elsewhere, but it does not meet the decisive requirement here: separate analytical copy/warehouse.

Question 20

An application requires the latest committed balance for a financial authorization. Which read path should it prefer?

Use the Aurora reader endpoint for eligible read-only connections.
Use an Aurora Global Database or supported cross-Region read replicas for regional read locality.
Copy analytical data to a separate warehouse or analytical replica.
Run the reporting workload against a read replica.
Read from the authoritative writer or another data path that provides the required current/strong consistency.

Correct Answer: E

Correct Answer

Answer E is correct because A replica with asynchronous lag is not appropriate when the decision requires the latest committed state. This directly meets the decisive requirement: authoritative writer/strongly consistent path.

Incorrect Answers

Answer A is incorrect because The reader endpoint distributes connections among available Aurora Replicas rather than sending them to the writer. It may be useful elsewhere, but it does not meet the decisive requirement here: authoritative writer/strongly consistent path.

Answer B is incorrect because Cross-Region replicas can place read capacity closer to users while preserving a designated writer Region. It may be useful elsewhere, but it does not meet the decisive requirement here: authoritative writer/strongly consistent path.

Answer C is incorrect because Separating heavy analytical scans prevents them from competing directly with the transactional writer workload. It may be useful elsewhere, but it does not meet the decisive requirement here: authoritative writer/strongly consistent path.

Question 21

Thousands of Lambda invocations each open a new RDS connection and the database exhausts its connection limit. Which AWS service should be added?

Keep transaction-dependent work on a compatible pinned/session-aware connection until the transaction completes.
Use Amazon RDS Proxy.
Reuse a bounded application connection pool.
Limit the application pool to a safe value derived from the database connection budget.
Implement resilient reconnection and retry logic with bounded backoff for failover events.

Correct Answer: B

Correct Answer

Answer B is correct because RDS Proxy pools and shares database connections and is especially useful for highly concurrent or serverless clients. This directly meets the decisive requirement: RDS Proxy.

Incorrect Answers

Answer A is incorrect because Transactional state cannot be arbitrarily moved among unrelated sessions while the transaction is active. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS Proxy.

Answer C is incorrect because Pooling avoids the cost and database pressure of establishing a new connection for every request. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS Proxy.

Answer D is incorrect because A client pool larger than the database can support simply moves the bottleneck into the database and increases failure risk. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS Proxy.

Answer E is incorrect because Clients must discard failed sessions and re-establish connections to the new healthy database endpoint after failover. It may be useful elsewhere, but it does not meet the decisive requirement here: RDS Proxy.

Question 22

A long-running application already uses a process-level database connection pool effectively. What should it do instead of opening a new TCP connection for every query?

Keep transaction-dependent work on a compatible pinned/session-aware connection until the transaction completes.
Reuse a bounded application connection pool.
Implement resilient reconnection and retry logic with bounded backoff for failover events.
Limit the application pool to a safe value derived from the database connection budget.
Use Amazon RDS Proxy.

Correct Answer: B

Correct Answer

Answer B is correct because Pooling avoids the cost and database pressure of establishing a new connection for every request. This directly meets the decisive requirement: reuse pooled connections.

Incorrect Answers

Answer C is incorrect because Clients must discard failed sessions and re-establish connections to the new healthy database endpoint after failover. It may be useful elsewhere, but it does not meet the decisive requirement here: reuse pooled connections.

Answer E is incorrect because RDS Proxy pools and shares database connections and is especially useful for highly concurrent or serverless clients. It may be useful elsewhere, but it does not meet the decisive requirement here: reuse pooled connections.

Question 23

A database fails over and applications keep stale broken connections for minutes. Which client behavior should be improved?

Keep transaction-dependent work on a compatible pinned/session-aware connection until the transaction completes.
Reuse a bounded application connection pool.
Implement resilient reconnection and retry logic with bounded backoff for failover events.
Use Amazon RDS Proxy.
Limit the application pool to a safe value derived from the database connection budget.

Correct Answer: C

Correct Answer

Answer C is correct because Clients must discard failed sessions and re-establish connections to the new healthy database endpoint after failover. This directly meets the decisive requirement: reconnect/retry with bounded backoff.

Incorrect Answers

Answer B is incorrect because Pooling avoids the cost and database pressure of establishing a new connection for every request. It may be useful elsewhere, but it does not meet the decisive requirement here: reconnect/retry with bounded backoff.

Answer D is incorrect because RDS Proxy pools and shares database connections and is especially useful for highly concurrent or serverless clients. It may be useful elsewhere, but it does not meet the decisive requirement here: reconnect/retry with bounded backoff.

Answer E is incorrect because A client pool larger than the database can support simply moves the bottleneck into the database and increases failure risk. It may be useful elsewhere, but it does not meet the decisive requirement here: reconnect/retry with bounded backoff.

Question 24

A transaction spans several SQL statements and must preserve session state until commit. What should the application avoid?

Keep transaction-dependent work on a compatible pinned/session-aware connection until the transaction completes.
Reuse a bounded application connection pool.
Implement resilient reconnection and retry logic with bounded backoff for failover events.
Use Amazon RDS Proxy.
Limit the application pool to a safe value derived from the database connection budget.

Correct Answer: A

Correct Answer

Answer A is correct because Transactional state cannot be arbitrarily moved among unrelated sessions while the transaction is active. This directly meets the decisive requirement: uncontrolled multiplexing across independent sessions.

Incorrect Answers

Question 25

The application pool allows 5,000 connections but the database safely supports only 1,000. Which configuration should be changed?

Reuse a bounded application connection pool.
Implement resilient reconnection and retry logic with bounded backoff for failover events.
Limit the application pool to a safe value derived from the database connection budget.
Use Amazon RDS Proxy.
Keep transaction-dependent work on a compatible pinned/session-aware connection until the transaction completes.

Correct Answer: C

Correct Answer

Answer C is correct because A client pool larger than the database can support simply moves the bottleneck into the database and increases failure risk. This directly meets the decisive requirement: cap pool below safe database limit.

Incorrect Answers

Answer A is incorrect because Pooling avoids the cost and database pressure of establishing a new connection for every request. It may be useful elsewhere, but it does not meet the decisive requirement here: cap pool below safe database limit.

Answer B is incorrect because Clients must discard failed sessions and re-establish connections to the new healthy database endpoint after failover. It may be useful elsewhere, but it does not meet the decisive requirement here: cap pool below safe database limit.

Answer E is incorrect because Transactional state cannot be arbitrarily moved among unrelated sessions while the transaction is active. It may be useful elsewhere, but it does not meet the decisive requirement here: cap pool below safe database limit.

Question 26

A product page reads the same catalog rows thousands of times per minute and updates only a few times per hour. Which caching pattern is appropriate?

Use ElastiCache as a shared low-latency session store when the application semantics fit.
Choose and understand the cache eviction policy.
Set an appropriate TTL on cached entries.
Use cache-aside: read the cache first, load a miss from the database, then populate the cache.
Design explicit cache invalidation or accept/document the allowed staleness window.

Correct Answer: D

Correct Answer

Answer D is correct because Cache-aside is well suited to read-heavy data where the application can tolerate explicit population and invalidation behavior. This directly meets the decisive requirement: cache-aside.

Incorrect Answers

Answer A is incorrect because A managed in-memory store can hold ephemeral session data independently of the lifecycle of individual web servers. It may be useful elsewhere, but it does not meet the decisive requirement here: cache-aside.

Answer B is incorrect because Eviction determines which entries are removed when the cache reaches its memory limit. It may be useful elsewhere, but it does not meet the decisive requirement here: cache-aside.

Answer C is incorrect because TTL bounds how long an item remains eligible to be served without refresh. It may be useful elsewhere, but it does not meet the decisive requirement here: cache-aside.

Answer E is incorrect because A cache can serve older data until it is invalidated, updated, or expires. It may be useful elsewhere, but it does not meet the decisive requirement here: cache-aside.

Question 27

A user profile cache sometimes shows data that changed seconds ago in the database. Which property must the design account for?

Use cache-aside: read the cache first, load a miss from the database, then populate the cache.
Choose and understand the cache eviction policy.
Use ElastiCache as a shared low-latency session store when the application semantics fit.
Set an appropriate TTL on cached entries.
Design explicit cache invalidation or accept/document the allowed staleness window.

Correct Answer: E

Correct Answer

Answer E is correct because A cache can serve older data until it is invalidated, updated, or expires. This directly meets the decisive requirement: staleness/invalidation.

Incorrect Answers

Answer A is incorrect because Cache-aside is well suited to read-heavy data where the application can tolerate explicit population and invalidation behavior. It may be useful elsewhere, but it does not meet the decisive requirement here: staleness/invalidation.

Answer C is incorrect because A managed in-memory store can hold ephemeral session data independently of the lifecycle of individual web servers. It may be useful elsewhere, but it does not meet the decisive requirement here: staleness/invalidation.

Answer D is incorrect because TTL bounds how long an item remains eligible to be served without refresh. It may be useful elsewhere, but it does not meet the decisive requirement here: staleness/invalidation.

Question 28

Price data must never remain cached longer than five minutes. Which cache setting is central to the requirement?

Design explicit cache invalidation or accept/document the allowed staleness window.
Use ElastiCache as a shared low-latency session store when the application semantics fit.
Choose and understand the cache eviction policy.
Use cache-aside: read the cache first, load a miss from the database, then populate the cache.
Set an appropriate TTL on cached entries.

Correct Answer: E

Correct Answer

Answer E is correct because TTL bounds how long an item remains eligible to be served without refresh. This directly meets the decisive requirement: TTL.

Incorrect Answers

Answer A is incorrect because A cache can serve older data until it is invalidated, updated, or expires. It may be useful elsewhere, but it does not meet the decisive requirement here: TTL.

Answer B is incorrect because A managed in-memory store can hold ephemeral session data independently of the lifecycle of individual web servers. It may be useful elsewhere, but it does not meet the decisive requirement here: TTL.

Answer C is incorrect because Eviction determines which entries are removed when the cache reaches its memory limit. It may be useful elsewhere, but it does not meet the decisive requirement here: TTL.

Answer D is incorrect because Cache-aside is well suited to read-heavy data where the application can tolerate explicit population and invalidation behavior. It may be useful elsewhere, but it does not meet the decisive requirement here: TTL.

Question 29

A cache is full and must remove entries to admit new values. Which behavior determines what data is discarded?

Set an appropriate TTL on cached entries.
Use ElastiCache as a shared low-latency session store when the application semantics fit.
Choose and understand the cache eviction policy.
Use cache-aside: read the cache first, load a miss from the database, then populate the cache.
Design explicit cache invalidation or accept/document the allowed staleness window.

Correct Answer: C

Correct Answer

Answer C is correct because Eviction determines which entries are removed when the cache reaches its memory limit. This directly meets the decisive requirement: eviction policy.

Incorrect Answers

Answer A is incorrect because TTL bounds how long an item remains eligible to be served without refresh. It may be useful elsewhere, but it does not meet the decisive requirement here: eviction policy.

Answer E is incorrect because A cache can serve older data until it is invalidated, updated, or expires. It may be useful elsewhere, but it does not meet the decisive requirement here: eviction policy.

Question 30

A web application stores short-lived login session state that must be shared across many stateless servers. Which use case fits ElastiCache?

Use cache-aside: read the cache first, load a miss from the database, then populate the cache.
Set an appropriate TTL on cached entries.
Choose and understand the cache eviction policy.
Use ElastiCache as a shared low-latency session store when the application semantics fit.
Design explicit cache invalidation or accept/document the allowed staleness window.

Correct Answer: D

Correct Answer

Answer D is correct because A managed in-memory store can hold ephemeral session data independently of the lifecycle of individual web servers. This directly meets the decisive requirement: in-memory session store.

Incorrect Answers

Answer B is incorrect because TTL bounds how long an item remains eligible to be served without refresh. It may be useful elsewhere, but it does not meet the decisive requirement here: in-memory session store.

Question 31

A service performs 100 times more reads than writes. Which capacity dimension should dominate initial performance testing?

Scale database compute because CPU is the saturated resource.
Increase or redesign the write I/O path rather than adding CPU alone.
Size and test read capacity for the dominant access pattern.
Increase memory or introduce an appropriate cache so more of the working set can be served without repeated storage reads.
Account for item size when estimating DynamoDB read/write capacity and network transfer.

Correct Answer: C

Correct Answer

Answer C is correct because A read-heavy workload should be validated against the read throughput and latency needed at peak demand. This directly meets the decisive requirement: read throughput.

Incorrect Answers

Answer A is incorrect because High database CPU with healthy storage metrics points to compute rather than storage as the immediate constraint. It may be useful elsewhere, but it does not meet the decisive requirement here: read throughput.

Answer B is incorrect because Write IOPS saturation is a storage bottleneck when compute has spare capacity. It may be useful elsewhere, but it does not meet the decisive requirement here: read throughput.

Answer D is incorrect because A working set much larger than memory can create avoidable I/O and latency. It may be useful elsewhere, but it does not meet the decisive requirement here: read throughput.

Answer E is incorrect because Larger DynamoDB items consume more capacity units and bytes per operation than small items. It may be useful elsewhere, but it does not meet the decisive requirement here: read throughput.

Question 32

A write-intensive ledger is saturating storage write IOPS while CPU remains low. Which capacity resource should be increased or redesigned?

Scale database compute because CPU is the saturated resource.
Size and test read capacity for the dominant access pattern.
Account for item size when estimating DynamoDB read/write capacity and network transfer.
Increase or redesign the write I/O path rather than adding CPU alone.
Increase memory or introduce an appropriate cache so more of the working set can be served without repeated storage reads.

Correct Answer: D

Correct Answer

Answer D is correct because Write IOPS saturation is a storage bottleneck when compute has spare capacity. This directly meets the decisive requirement: write IOPS/storage path.

Incorrect Answers

Answer B is incorrect because A read-heavy workload should be validated against the read throughput and latency needed at peak demand. It may be useful elsewhere, but it does not meet the decisive requirement here: write IOPS/storage path.

Answer C is incorrect because Larger DynamoDB items consume more capacity units and bytes per operation than small items. It may be useful elsewhere, but it does not meet the decisive requirement here: write IOPS/storage path.

Answer E is incorrect because A working set much larger than memory can create avoidable I/O and latency. It may be useful elsewhere, but it does not meet the decisive requirement here: write IOPS/storage path.

Question 33

DynamoDB items grow from 1 KB to 300 KB with the same request rate. What capacity effect must be considered?

Scale database compute because CPU is the saturated resource.
Increase or redesign the write I/O path rather than adding CPU alone.
Increase memory or introduce an appropriate cache so more of the working set can be served without repeated storage reads.
Size and test read capacity for the dominant access pattern.
Account for item size when estimating DynamoDB read/write capacity and network transfer.

Correct Answer: E

Correct Answer

Answer E is correct because Larger DynamoDB items consume more capacity units and bytes per operation than small items. This directly meets the decisive requirement: capacity consumed per request rises with item size.

Incorrect Answers

Answer C is incorrect because A working set much larger than memory can create avoidable I/O and latency. It may be useful elsewhere, but it does not meet the decisive requirement here: capacity consumed per request rises with item size.

Answer D is incorrect because A read-heavy workload should be validated against the read throughput and latency needed at peak demand. It may be useful elsewhere, but it does not meet the decisive requirement here: capacity consumed per request rises with item size.

Question 34

An RDS instance shows high CPU but low storage latency and the working set fits in memory. Which resource is most likely limiting?

Increase or redesign the write I/O path rather than adding CPU alone.
Increase memory or introduce an appropriate cache so more of the working set can be served without repeated storage reads.
Account for item size when estimating DynamoDB read/write capacity and network transfer.
Scale database compute because CPU is the saturated resource.
Size and test read capacity for the dominant access pattern.

Correct Answer: D

Correct Answer

Answer D is correct because High database CPU with healthy storage metrics points to compute rather than storage as the immediate constraint. This directly meets the decisive requirement: compute capacity.

Incorrect Answers

Answer A is incorrect because Write IOPS saturation is a storage bottleneck when compute has spare capacity. It may be useful elsewhere, but it does not meet the decisive requirement here: compute capacity.

Answer B is incorrect because A working set much larger than memory can create avoidable I/O and latency. It may be useful elsewhere, but it does not meet the decisive requirement here: compute capacity.

Answer E is incorrect because A read-heavy workload should be validated against the read throughput and latency needed at peak demand. It may be useful elsewhere, but it does not meet the decisive requirement here: compute capacity.

Question 35

A database repeatedly reads a 50 GB working set from storage because only 8 GB of memory is available. Which resource change can improve performance?

Size and test read capacity for the dominant access pattern.
Account for item size when estimating DynamoDB read/write capacity and network transfer.
Increase or redesign the write I/O path rather than adding CPU alone.
Increase memory or introduce an appropriate cache so more of the working set can be served without repeated storage reads.
Scale database compute because CPU is the saturated resource.

Correct Answer: D

Correct Answer

Answer D is correct because A working set much larger than memory can create avoidable I/O and latency. This directly meets the decisive requirement: more memory or cache.

Incorrect Answers

Answer A is incorrect because A read-heavy workload should be validated against the read throughput and latency needed at peak demand. It may be useful elsewhere, but it does not meet the decisive requirement here: more memory or cache.

Answer B is incorrect because Larger DynamoDB items consume more capacity units and bytes per operation than small items. It may be useful elsewhere, but it does not meet the decisive requirement here: more memory or cache.

Answer C is incorrect because Write IOPS saturation is a storage bottleneck when compute has spare capacity. It may be useful elsewhere, but it does not meet the decisive requirement here: more memory or cache.

Answer E is incorrect because High database CPU with healthy storage metrics points to compute rather than storage as the immediate constraint. It may be useful elsewhere, but it does not meet the decisive requirement here: more memory or cache.

Question 36

A company wants complex analytical SQL over years of fact data without running those scans on the OLTP database. Which service should host the analytical workload?

Replicate changes through a streaming or change-data-capture pipeline into the analytical store.
Separate the analytical workload because long scans can consume CPU, I/O, locks, or cache needed by transactions.
Use Amazon Redshift for the analytical warehouse.
Use a read replica for reporting if the engine supports it and the lag is acceptable.
Use Amazon OpenSearch Service for the search index.

Correct Answer: C

Correct Answer

Answer C is correct because Redshift is designed for large-scale analytical SQL and separates warehouse workloads from the transactional database. This directly meets the decisive requirement: Redshift.

Incorrect Answers

Answer A is incorrect because CDC or streaming keeps an independent analytical copy current without forcing all analytics onto the OLTP writer. It may be useful elsewhere, but it does not meet the decisive requirement here: Redshift.

Answer B is incorrect because Keeping analytics on the primary can cause direct resource contention and unpredictable OLTP latency. It may be useful elsewhere, but it does not meet the decisive requirement here: Redshift.

Answer D is incorrect because A replica can offload read-only reporting with less architectural complexity than a separate warehouse when requirements are modest. It may be useful elsewhere, but it does not meet the decisive requirement here: Redshift.

Answer E is incorrect because OpenSearch provides full-text indexing, relevance search, and aggregations that differ from transactional database access patterns. It may be useful elsewhere, but it does not meet the decisive requirement here: Redshift.

Question 37

A product catalog needs full-text search, relevance ranking, and faceted queries that are awkward in the transactional schema. Which secondary data store is appropriate?

Use Amazon OpenSearch Service for the search index.
Use Amazon Redshift for the analytical warehouse.
Use a read replica for reporting if the engine supports it and the lag is acceptable.
Separate the analytical workload because long scans can consume CPU, I/O, locks, or cache needed by transactions.
Replicate changes through a streaming or change-data-capture pipeline into the analytical store.

Correct Answer: A

Correct Answer

Answer A is correct because OpenSearch provides full-text indexing, relevance search, and aggregations that differ from transactional database access patterns. This directly meets the decisive requirement: OpenSearch.

Incorrect Answers

Answer B is incorrect because Redshift is designed for large-scale analytical SQL and separates warehouse workloads from the transactional database. It may be useful elsewhere, but it does not meet the decisive requirement here: OpenSearch.

Answer C is incorrect because A replica can offload read-only reporting with less architectural complexity than a separate warehouse when requirements are modest. It may be useful elsewhere, but it does not meet the decisive requirement here: OpenSearch.

Answer D is incorrect because Keeping analytics on the primary can cause direct resource contention and unpredictable OLTP latency. It may be useful elsewhere, but it does not meet the decisive requirement here: OpenSearch.

Answer E is incorrect because CDC or streaming keeps an independent analytical copy current without forcing all analytics onto the OLTP writer. It may be useful elsewhere, but it does not meet the decisive requirement here: OpenSearch.

Question 38

An operational database must feed analytics with changes within seconds. Which architecture pattern is appropriate?

Use Amazon Redshift for the analytical warehouse.
Use a read replica for reporting if the engine supports it and the lag is acceptable.
Use Amazon OpenSearch Service for the search index.
Separate the analytical workload because long scans can consume CPU, I/O, locks, or cache needed by transactions.
Replicate changes through a streaming or change-data-capture pipeline into the analytical store.

Correct Answer: E

Correct Answer

Answer E is correct because CDC or streaming keeps an independent analytical copy current without forcing all analytics onto the OLTP writer. This directly meets the decisive requirement: stream or CDC changes to analytical store.

Incorrect Answers

Answer A is incorrect because Redshift is designed for large-scale analytical SQL and separates warehouse workloads from the transactional database. It may be useful elsewhere, but it does not meet the decisive requirement here: stream or CDC changes to analytical store.

Answer B is incorrect because A replica can offload read-only reporting with less architectural complexity than a separate warehouse when requirements are modest. It may be useful elsewhere, but it does not meet the decisive requirement here: stream or CDC changes to analytical store.

Answer C is incorrect because OpenSearch provides full-text indexing, relevance search, and aggregations that differ from transactional database access patterns. It may be useful elsewhere, but it does not meet the decisive requirement here: stream or CDC changes to analytical store.

Question 39

Finance runs daily reporting queries and can tolerate replica lag, but the OLTP writer must remain isolated. Which low-complexity option is suitable?

Separate the analytical workload because long scans can consume CPU, I/O, locks, or cache needed by transactions.
Replicate changes through a streaming or change-data-capture pipeline into the analytical store.
Use Amazon OpenSearch Service for the search index.
Use a read replica for reporting if the engine supports it and the lag is acceptable.
Use Amazon Redshift for the analytical warehouse.

Correct Answer: D

Correct Answer

Answer D is correct because A replica can offload read-only reporting with less architectural complexity than a separate warehouse when requirements are modest. This directly meets the decisive requirement: read replica.

Incorrect Answers

Answer A is incorrect because Keeping analytics on the primary can cause direct resource contention and unpredictable OLTP latency. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Answer B is incorrect because CDC or streaming keeps an independent analytical copy current without forcing all analytics onto the OLTP writer. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Answer E is incorrect because Redshift is designed for large-scale analytical SQL and separates warehouse workloads from the transactional database. It may be useful elsewhere, but it does not meet the decisive requirement here: read replica.

Question 40

An application team proposes running all historical analytics directly against the primary production database because it already contains the data. What is the main architectural concern?

Separate the analytical workload because long scans can consume CPU, I/O, locks, or cache needed by transactions.
Use Amazon OpenSearch Service for the search index.
Replicate changes through a streaming or change-data-capture pipeline into the analytical store.
Use Amazon Redshift for the analytical warehouse.
Use a read replica for reporting if the engine supports it and the lag is acceptable.

Correct Answer: A

Correct Answer

Answer A is correct because Keeping analytics on the primary can cause direct resource contention and unpredictable OLTP latency. This directly meets the decisive requirement: resource contention with transactions.

Incorrect Answers

Answer B is incorrect because OpenSearch provides full-text indexing, relevance search, and aggregations that differ from transactional database access patterns. It may be useful elsewhere, but it does not meet the decisive requirement here: resource contention with transactions.

Answer C is incorrect because CDC or streaming keeps an independent analytical copy current without forcing all analytics onto the OLTP writer. It may be useful elsewhere, but it does not meet the decisive requirement here: resource contention with transactions.

Answer D is incorrect because Redshift is designed for large-scale analytical SQL and separates warehouse workloads from the transactional database. It may be useful elsewhere, but it does not meet the decisive requirement here: resource contention with transactions.

Answer E is incorrect because A replica can offload read-only reporting with less architectural complexity than a separate warehouse when requirements are modest. It may be useful elsewhere, but it does not meet the decisive requirement here: resource contention with transactions.

Related posts: