Knowledge + Practice

CCNA Db Mgmt Ops Questions

75 of 300 questions · Page 3/4 · Db Mgmt Ops topic · Answers revealed

Practice these questions Exam hub All questions

151

MCQhard

A financial services company runs a production Amazon Aurora MySQL database cluster (1 writer, 2 readers) in us-east-1. The database stores critical trading data. The company's disaster recovery policy requires an RPO of 5 seconds and an RTO of 1 minute for a regional failure. The current setup does not include any cross-region replication. The database is 5 TB in size. The operations team needs to implement a solution that meets the DR requirements with minimal cost and operational overhead. Which solution should the team implement?

A.Enable automated backups and configure cross-Region snapshot copy. Use point-in-time recovery in the secondary region.

B.Deploy an Aurora Global Database with a secondary cluster in us-west-2. Configure the secondary cluster as a failover target.

C.Use AWS Database Migration Service (DMS) to continuously replicate changes to an Aurora cluster in us-west-2.

D.Create cross-Region read replicas in us-west-2 and set up a replication channel. In disaster, promote a replica.

AnswerB

Aurora Global Database provides replication across regions with typical RPO of seconds and RTO of minutes. It is the most appropriate solution for low RPO/RTO with minimal overhead.

Why this answer

Aurora Global Database provides replication across regions with typical RPO of seconds and RTO of minutes. It is the most appropriate solution for low RPO/RTO with minimal overhead. Option B is correct.

Option A (automated backups) have an RPO of 5 minutes, which exceeds the required 5 seconds. Option C (DMS) adds cost and complexity. Option D (cross-Region read replicas) can have replication lag exceeding 5 seconds.

Practice this question →

152

MCQeasy

A company is using Amazon DynamoDB with on-demand capacity for a serverless web application. The application experiences occasional throttling. The DynamoDB table has a simple primary key (partition key only). The throttled requests are related to a small number of partition keys. What is the MOST likely cause?

A.On-demand capacity has a per-partition throughput limit that is too low.

B.The partition key design leads to uneven access patterns, causing a hot partition.

C.The table uses a composite primary key, which limits throughput.

D.The table's read/write capacity mode is set to provisioned instead of on-demand.

AnswerB

A hot partition exceeds the partition's throughput limit, causing throttling.

Why this answer

Throttling on a few partition keys indicates a hot partition. Option A is wrong because on-demand capacity handles overall traffic, but partition-level limits still apply. Option C is wrong because a simple primary key is fine, but the data distribution is the issue.

Option D is wrong because on-demand capacity automatically scales, but not per partition beyond the limit.

Practice this question →

153

MCQmedium

A company runs an Amazon Redshift cluster with three nodes. The data warehouse team notices that some queries are slow due to high disk usage. The cluster has reached 80% storage capacity. What is the MOST cost-effective way to increase storage without interrupting operations?

A.Add one more node of the same type to the cluster.

B.Use classic resize to change to a node type with larger storage.

C.Perform an elastic resize to change to a node type with larger storage per node.

D.Create a second cluster and use Redshift Spectrum to offload queries.

AnswerC

Elastic resize completes in minutes and minimizes downtime.

Why this answer

Amazon Redshift's elastic resize operation allows you to change the node type to one with larger storage per node without downtime, and it completes in minutes. This is the most cost-effective approach for a cluster at 80% capacity, as it avoids the overhead of provisioning additional nodes or the longer downtime associated with classic resize, while directly addressing the high disk usage by increasing per-node storage.

Exam trap

The trap here is that candidates often confuse elastic resize with classic resize, assuming both require downtime, or they incorrectly think adding a node of the same type is a simple online operation, when in fact Redshift does not support online addition of nodes without a resize operation.

How to eliminate wrong answers

Option A is wrong because adding a node of the same type increases compute and storage capacity, but it requires a classic resize (which involves downtime) or an elastic resize (which only supports changing node type, not adding nodes of the same type). Option B is wrong because classic resize requires the cluster to be read-only during the operation and can take hours, causing significant operational interruption, which violates the requirement to not interrupt operations. Option D is wrong because creating a second cluster and using Redshift Spectrum offloads queries to Amazon S3, but it does not increase the storage capacity of the existing cluster; Spectrum is for querying external data, not for expanding local disk space, and it adds complexity and cost without solving the high disk usage issue.

Practice this question →

154

MCQmedium

A company uses Amazon ElastiCache for Redis as a caching layer for its e-commerce application. Recently, the cache hit ratio has dropped significantly, causing increased database load. The operations team needs to identify which cache keys are being evicted. What should they do?

A.Monitor the 'Evictions' metric in Amazon CloudWatch for the ElastiCache cluster.

B.Check the ElastiCache event history for eviction events.

C.Enable the 'INFO' command output to be logged to CloudWatch Logs.

D.Enable the Redis slow-log to capture eviction commands.

AnswerA

CloudWatch provides the evictions metric which tracks the number of evicted keys.

Why this answer

The 'Evictions' metric in Amazon CloudWatch directly reports the number of keys evicted from the ElastiCache for Redis cluster due to memory pressure. A drop in cache hit ratio often correlates with increased evictions, and monitoring this metric allows the operations team to identify the rate at which keys are being removed. This is the standard, built-in way to observe eviction activity without additional configuration.

Exam trap

The trap here is that candidates may confuse event history (cluster-level events) with data-level operations (key evictions), or assume that logging the INFO command or slow-log would capture evictions, when in fact evictions are not commands and are best monitored via CloudWatch metrics.

How to eliminate wrong answers

Option B is wrong because ElastiCache event history records cluster-level events (e.g., node replacements, scaling, failover) but does not log individual key evictions; evictions are not emitted as events. Option C is wrong because enabling the Redis 'INFO' command output to CloudWatch Logs provides a snapshot of server statistics (including eviction counts) but is not a real-time metric and requires parsing logs; CloudWatch metrics like 'Evictions' are more direct and actionable. Option D is wrong because the Redis slow-log captures commands that exceed a specified execution time threshold, not eviction events; evictions are not commands but automatic memory management actions, so they never appear in the slow-log.

Practice this question →

155

Multi-Selecthard

Which THREE factors should be considered when choosing between Amazon DynamoDB and Amazon RDS for a new application? (Choose three.)

Select 3 answers

A.Need for ACID transactions across multiple rows

B.Need for complex join queries across multiple tables

C.Need for automatic failover in multiple AWS Regions

D.Requirement for flexible schema with document data

E.Data size exceeding 10 TB

AnswersA, B, D

RDS supports full ACID; DynamoDB supports transactional APIs but limited.

Why this answer

Amazon RDS supports ACID transactions across multiple rows using traditional SQL databases like MySQL or PostgreSQL, which is essential for applications requiring strict consistency (e.g., financial systems). Amazon DynamoDB, while supporting ACID transactions via the TransactGetItems and TransactWriteItems APIs, is optimized for single-item or limited multi-item operations and may not perform as well for complex multi-row transactional workloads. The need for ACID across multiple rows is a key differentiator favoring RDS.

Exam trap

The trap here is that candidates assume automatic failover across multiple Regions is unique to one service, but both DynamoDB (global tables) and RDS (cross-Region read replicas with manual promotion) can achieve this, making it a non-differentiating factor.

Practice this question →

156

MCQmedium

A company is using Amazon Redshift for data warehousing. The database administrator needs to identify which queries are consuming the most resources. Which system view should be queried?

A.SVV_TABLES

B.STV_RECENTS

C.STL_LOAD_ERRORS

D.STL_DDLTEXT

AnswerB

STV_RECENTS shows active and recent queries, including their resource usage.

Why this answer

The STV_RECENTS system view in Amazon Redshift provides a list of currently running and recently completed queries, including their process IDs, user names, and execution status. This makes it the correct choice for identifying which queries are consuming the most resources at the moment, as it directly reflects active and recent workload.

Exam trap

The trap here is that candidates confuse system views for metadata (SVV_TABLES) or error logging (STL_LOAD_ERRORS) with those that track query execution and resource usage, leading them to overlook STV_RECENTS as the direct source for active query monitoring.

How to eliminate wrong answers

Option A is wrong because SVV_TABLES is a system view that lists tables and their metadata (like schema, table name, and table type), not query resource consumption. Option C is wrong because STL_LOAD_ERRORS logs errors that occur during COPY or INSERT operations, focusing on data load failures rather than general query resource usage. Option D is wrong because STL_DDLTEXT captures the text of DDL statements (e.g., CREATE, ALTER) that have been executed, not runtime resource consumption of queries.

Practice this question →

157

Multi-Selectmedium

Which THREE of the following are best practices for managing Amazon DynamoDB tables with provisioned throughput?

Select 3 answers

A.Create a global secondary index on every attribute to support any query pattern.

B.Split hot partitions manually to distribute write traffic.

C.Use a composite key design with a sort key to enable efficient querying.

D.Use DynamoDB Auto Scaling to adjust read/write capacity based on traffic.

E.Enable DynamoDB Accelerator (DAX) to improve read performance for frequently accessed items.

AnswersC, D, E

Composite keys allow efficient range queries and data organization.

Why this answer

Options C, D, and E are correct. Option A is incorrect because creating a global secondary index on every attribute incurs additional costs and storage, and not every query pattern requires an index; GSIs should be designed based on actual access patterns. Option B is incorrect because DynamoDB automatically manages partitions; manual partition splitting is not supported.

Option C is correct because a composite key design (partition key and sort key) enables efficient querying using the sort key for filtering and sorting. Option D is correct because DynamoDB Auto Scaling automatically adjusts read/write capacity based on actual traffic, helping to maintain performance without manual intervention. Option E is correct because DynamoDB Accelerator (DAX) provides in-memory caching for frequently accessed items, reducing read latency and improving performance.

Practice this question →

158

MCQhard

A company is migrating an on-premises Microsoft SQL Server database to Amazon RDS for SQL Server. The database uses SQL Server Agent jobs, custom CLR assemblies, and cross-database queries. Which of the following will require modification before migration?

A.Cross-database queries

B.Custom CLR assemblies

C.Stored procedures that use dynamic SQL

D.SQL Server Agent jobs

AnswerD

SQL Server Agent is not available in RDS; jobs require alternative solutions.

Why this answer

SQL Server Agent jobs are not supported in Amazon RDS for SQL Server because RDS is a managed service that does not provide access to the underlying operating system or the SQL Server Agent service. To migrate job scheduling, you must use alternatives such as AWS Database Migration Service (DMS) tasks, AWS Lambda, or Amazon RDS for SQL Server native scheduling via stored procedures and Windows Task Scheduler on an EC2 instance.

Exam trap

The trap here is that candidates assume SQL Server Agent jobs are fully supported in RDS because RDS for SQL Server includes the SQL Server engine, but they overlook that Agent is a separate Windows service that RDS does not expose, requiring a workaround for job scheduling.

How to eliminate wrong answers

Option A is wrong because cross-database queries are supported in Amazon RDS for SQL Server as long as the databases are within the same RDS instance; no modification is required for queries that reference tables in other databases on the same instance. Option B is wrong because custom CLR assemblies are supported in Amazon RDS for SQL Server, provided they are signed with a certificate or asymmetric key and the CLR integration is enabled via the rds_custom_clr option group setting. Option C is wrong because stored procedures that use dynamic SQL are fully supported in Amazon RDS for SQL Server, as dynamic SQL execution is a core T-SQL feature that does not require any special configuration or modification.

Practice this question →

159

Multi-Selecthard

A database specialist is troubleshooting a performance issue on an Amazon Aurora MySQL DB cluster. The cluster has a primary instance and two Aurora Replicas. The application is experiencing high read latency. Which THREE actions should the specialist take to identify the cause?

Select 3 answers

A.Check the Write Latency metric on the primary instance.

B.Increase the instance size of the Aurora Replicas.

C.Check the Performance Insights for the Aurora Replicas to identify long-running queries.

D.Monitor the Aurora ReplicaLag metric in Amazon CloudWatch.

E.Evaluate the DB Connection Count metric to see if the replicas are overloaded.

AnswersC, D, E

Long-running queries can cause read latency.

Why this answer

Performance Insights provides a detailed view of database load, including wait events, SQL queries, and host-level metrics, which can pinpoint long-running queries causing high read latency on the Aurora Replicas. By analyzing the top SQL statements and their execution plans, the specialist can identify inefficient queries or contention issues that degrade read performance.

Exam trap

The trap here is that candidates may confuse write latency on the primary with read latency on replicas, or assume scaling the replica size is a diagnostic step rather than a remediation, leading them to select Option A or B instead of focusing on monitoring and analysis metrics.

Practice this question →

160

MCQeasy

A database administrator needs to audit all SQL queries executed on an Amazon RDS for PostgreSQL instance, including SELECT statements. Which two most efficient ways to achieve this?

A.Enable AWS CloudTrail for the RDS instance.

B.Install the pgAudit extension and configure it to log all statements.

C.Configure the DB instance to export logs to Amazon CloudWatch Logs.

D.Enable Enhanced Monitoring for the DB instance.

AnswerB, C

Correct. The pgAudit extension is the native way to log all SQL statements, including SELECTs, on PostgreSQL. It provides granular auditing.

Why this answer

The pgAudit extension for PostgreSQL can be installed on RDS and configured to log all SQL statements, including SELECT queries. Option C is also correct because configuring the DB instance to export logs to Amazon CloudWatch Logs, when combined with enabling statement logging via the RDS parameter group (e.g., setting log_statement = 'all'), allows centralized log management and analysis, effectively auditing all queries. Option A is wrong because AWS CloudTrail does not capture SQL query content; it logs API actions on the RDS instance itself.

Option D is wrong because Enhanced Monitoring provides OS-level metrics, not SQL query logs. Therefore, both B and C are correct.

Exam trap

Some candidates may think CloudTrail logs SQL queries, but it only records API actions. Others may overlook that exporting logs to CloudWatch alone does not capture query content unless statement logging is enabled in the parameter group.

Practice this question →

161

MCQeasy

A company is migrating its on-premises Oracle database to Amazon RDS for Oracle. The database is 2 TB in size and contains both transactional and analytical workloads. The company uses Oracle Data Pump for migration. During the migration, the database specialist notices that the export job on the source database is taking longer than expected and is generating significant I/O, impacting production performance. The company needs to minimize the impact on the source database while completing the migration within a 48-hour window. The source database is currently running on a dedicated server with sufficient CPU and memory. Which course of action should the database specialist take?

A.Use AWS Database Migration Service (AWS DMS) with full load and change data capture (CDC) to migrate the database.

B.Use Data Pump to export to a networked file system (NFS) mount point to offload storage I/O to the network.

C.Use Oracle Data Pump in network_link mode to directly import the data from the source to the target RDS instance without intermediate files.

D.Increase the parallel workers for the Data Pump export job to 8 to speed up the export and reduce the time window.

AnswerC

Correct: This minimizes source I/O as data is transferred directly.

Why this answer

Oracle Data Pump's network_link mode allows the source database to export data directly to the target RDS for Oracle instance over a database link, bypassing the need to write intermediate dump files to disk. This eliminates the I/O overhead on the source server's storage subsystem, which is the primary cause of the production performance impact, while still completing the migration within the 48-hour window.

Exam trap

The trap here is that candidates often assume increasing parallelism (Option D) is always the best way to speed up a Data Pump job, but they overlook that the primary issue is I/O impact on production, not throughput, and that network_link mode eliminates the I/O bottleneck entirely.

How to eliminate wrong answers

Option A is wrong because AWS DMS with full load and CDC would still require reading the source database, which can generate significant I/O and impact production performance; additionally, DMS may not support all Oracle-specific features or data types that Data Pump handles natively. Option B is wrong because exporting to an NFS mount point still writes dump files to a networked file system, which does not eliminate the I/O on the source server's local storage and can introduce network latency and additional overhead. Option D is wrong because increasing parallel workers for the Data Pump export job would increase I/O and CPU consumption on the source database, exacerbating the production performance impact rather than minimizing it.

Practice this question →

162

MCQhard

A database administrator is troubleshooting an Amazon RDS for PostgreSQL DB instance that is experiencing high CPU utilization. The administrator runs the following query to find the current running queries: SELECT pid, now() - pg_stat_activity.query_start AS duration, query, state FROM pg_stat_activity WHERE state = 'active'; The output shows a high number of queries with a state of 'active' and durations exceeding several minutes. What should the administrator do FIRST to reduce CPU utilization?

A.Modify the max_connections parameter to limit concurrent sessions.

B.Scale up the DB instance to a larger instance class.

C.Use pg_terminate_backend to terminate the long-running queries.

D.Enable pg_stat_statements to collect query performance data.

AnswerC

Terminating long-running queries immediately reduces CPU usage.

Why this answer

The immediate cause of high CPU utilization is the long-running active queries consuming resources. Using pg_terminate_backend to terminate these queries will quickly free up CPU cycles, providing immediate relief. This is the first troubleshooting step before making configuration changes or scaling, as it directly addresses the symptom shown in the pg_stat_activity output.

Exam trap

The trap here is that candidates may choose to scale up or adjust parameters first, overlooking that the immediate problem is the active long-running queries, which can be resolved quickly with pg_terminate_backend without incurring cost or configuration changes.

How to eliminate wrong answers

Option A is wrong because modifying max_connections limits the number of concurrent sessions but does not address the existing long-running queries that are already consuming CPU; it may also cause application errors if connections are rejected. Option B is wrong because scaling up the DB instance to a larger class is a reactive and costly measure that does not resolve the root cause of inefficient queries; it should only be considered after optimizing queries or confirming a sustained workload increase. Option D is wrong because enabling pg_stat_statements collects query performance data for analysis but does not reduce current CPU utilization; it is a diagnostic tool for future optimization, not an immediate fix.

Practice this question →

163

Multi-Selecthard

Which THREE factors should be considered when planning a cross-Region disaster recovery strategy for an Amazon Aurora MySQL DB cluster? (Choose 3.)

Select 3 answers

A.Provisioned IOPS for the primary cluster.

B.Network latency between source and target Regions.

C.Storage costs in the target Region.

D.Failover priority of reader instances.

E.Recovery Point Objective (RPO) requirements.

AnswersB, C, E

Latency affects replication lag and RTO.

Why this answer

Network latency between source and target Regions directly impacts the replication lag for Amazon Aurora Global Database. Aurora uses asynchronous replication across Regions, and higher latency increases the delay between committing a transaction on the primary cluster and it being available on the secondary cluster, which affects the achievable Recovery Point Objective (RPO).

Exam trap

The trap here is that candidates confuse local failover mechanisms (reader instance priority) with cross-Region disaster recovery, or mistakenly think that provisioned IOPS or storage costs in the target Region are primary considerations when RPO and replication lag are the critical factors.

Practice this question →

164

MCQhard

A media streaming company uses Amazon DynamoDB as the primary data store for user session information. The table has a partition key of 'user_id' (String) and a sort key of 'session_start_time' (Number). The table is provisioned with 5000 read capacity units (RCU) and 5000 write capacity units (WCU). Over the past week, the company noticed occasional throttling on writes during peak hours, even though the average consumed WCU is only 3500. The DynamoDB table also experiences hot partitions. The development team is considering using DynamoDB Accelerator (DAX) to cache read-heavy workloads, but the issue is with writes. The database administrator needs to resolve the write throttling and hot partitions. Which action should the administrator take?

A.Enable DynamoDB adaptive capacity to automatically handle uneven access patterns.

B.Deploy DAX to cache frequent writes and reduce write load.

C.Reduce the provisioned WCU to match the average consumption to save costs.

D.Enable DynamoDB auto scaling for write capacity to scale based on demand.

AnswerA

Adaptive capacity automatically adjusts throughput to handle uneven access patterns, helping with hot partitions and write throttling.

Why this answer

Adaptive capacity automatically adjusts throughput to handle uneven access patterns, helping with hot partitions. Option B is wrong because DAX caches reads, not writes; it does not help with write throttling. Option C is wrong because reducing WCU would worsen throttling, not resolve it.

Option D is wrong because enabling auto scaling adjusts capacity but does not directly address hot partitions; it may help with overall throughput but not uneven distribution.

Practice this question →

165

Multi-Selecthard

A company uses Amazon DynamoDB with auto scaling enabled. During a flash sale, write traffic spikes and some requests receive ThrottlingException errors. The table's write capacity auto scaling reaches the maximum configured capacity but still throttles. Which TWO actions should resolve the throttling? (Choose TWO.)

Select 2 answers

A.Increase the maximum write capacity setting for the auto scaling policy.

B.Review the partition key design to ensure even access distribution.

C.Add a DynamoDB Accelerator (DAX) cluster to cache writes.

D.Enable DynamoDB Streams to offload write traffic.

E.Disable auto scaling and use On-Demand capacity mode.

AnswersA, B

Allows auto scaling to scale higher.

Why this answer

Increases the maximum write capacity setting, allowing the auto scaling policy to scale higher and handle the increased traffic. Option B addresses hot partitions by reviewing the partition key design to ensure even distribution of writes, which reduces throttling even if capacity is sufficient. Option C (DAX) is a caching layer for reads, not writes, so it does not resolve write throttling.

Option D (DynamoDB Streams) captures changes but does not offload write traffic; it's a separate feature for streaming data. Option E (On-Demand mode) would help, but the question asks to resolve throttling with auto scaling enabled; On-Demand is an alternative, not a fix within auto scaling, but it's not the best choice given the context.

Practice this question →

166

MCQeasy

A company needs to ensure that all changes to an Amazon RDS DB instance's security group are logged for auditing purposes. Which AWS service should be enabled?

A.Amazon RDS Event Subscriptions

B.AWS CloudTrail

C.AWS Config

D.VPC Flow Logs

AnswerB

CloudTrail logs all API actions, including security group modifications.

Why this answer

AWS CloudTrail is the correct service because it records API calls made to the AWS environment, including actions like AuthorizeSecurityGroupIngress and RevokeSecurityGroupEgress that modify security group rules. When enabled, CloudTrail logs all changes to Amazon RDS security groups as management events, providing an immutable audit trail that can be stored in Amazon S3 for compliance and auditing purposes.

Exam trap

The trap here is that candidates confuse AWS Config's configuration tracking with CloudTrail's API logging, not realizing that Config evaluates resource compliance but does not capture the who, what, and when of API calls that change security groups.

How to eliminate wrong answers

Option A is wrong because Amazon RDS Event Subscriptions notify you about database instance events (e.g., failover, backup completion) but do not log or record API-level changes to security group configurations. Option C is wrong because AWS Config evaluates resource configurations against desired policies and tracks configuration changes, but it does not provide the detailed API call logs required for auditing who made the change and when; it relies on CloudTrail for that data. Option D is wrong because VPC Flow Logs capture IP traffic metadata (source/destination IP, ports, protocol) at the network interface level, not management-plane API calls that modify security group rules.

Practice this question →

167

Matchingmedium

Match each AWS database feature to its description.

Drag a concept onto its matching description — or click a concept then click the description.

Concepts

Matches

Synchronous standby replica for high availability

Asynchronous read-only copy for scaling reads

Automatic adjustment of Aurora replica count

In-memory cache for DynamoDB with microsecond latency

Connection pooling for RDS to handle Lambda bursts

Why these pairings

Multi-AZ is for high availability and failover; Read Replicas offload read traffic; Auto Scaling adjusts capacity; Point-in-Time Recovery enables granular restores. Common confusions involve swapping Multi-AZ and Read Replicas definitions.

Practice this question →

168

MCQeasy

A database administrator needs to monitor the number of connections to an Amazon RDS for PostgreSQL DB instance. Which Amazon CloudWatch metric should be used?

A.DatabaseConnections

B.CPUUtilization

C.WriteLatency

D.FreeableMemory

AnswerA

This metric tracks the number of connections to the DB instance.

Why this answer

The DatabaseConnections metric in Amazon CloudWatch directly reports the number of client connections to the RDS for PostgreSQL DB instance. This metric is published by the RDS service and reflects the count of active connections as seen by the database engine, making it the correct choice for monitoring connection counts.

Exam trap

The trap here is that candidates might confuse CPUUtilization or memory metrics with connection monitoring, but AWS explicitly provides the DatabaseConnections metric for this purpose, and the exam tests your knowledge of which specific CloudWatch metric maps to which operational need.

How to eliminate wrong answers

Option B (CPUUtilization) is wrong because it measures the percentage of CPU used by the instance, not the number of connections. Option C (WriteLatency) is wrong because it tracks the average time for write operations to complete, which is a performance metric unrelated to connection count. Option D (FreeableMemory) is wrong because it indicates the amount of available RAM on the instance, not the number of active connections.

Practice this question →

169

MCQhard

A company runs a production Amazon RDS for MySQL Multi-AZ DB instance with 10 TB of storage. The application team notices that database write latency has increased from 5 ms to 80 ms over the past week. The DB instance is a db.r5.8xlarge with 500 GB of gp2 storage. CloudWatch metrics show WriteIOPS averaging 10,000, WriteThroughput at 50 MB/s, and BurstBalance decreasing from 100% to 20% over the same period. The DB instance has no reserved IOPS. The database workload is write-heavy with frequent small transactions. The company needs to resolve the high write latency while minimizing costs and downtime. Which solution should the database specialist recommend?

A.Create a read replica to offload write traffic to the replica and use it for failover.

B.Upgrade the DB instance to a db.r5.16xlarge with more vCPUs and memory to handle higher write throughput.

C.Enable Performance Insights to identify slow queries and add database-level caching to reduce write load.

D.Modify the DB instance to increase the allocated storage to 1 TB using gp2, allowing the volume to operate with higher baseline IOPS and burst credits.

AnswerD

Increasing volume size boosts baseline IOPS and burst balance, resolving the bottleneck without changing instance type.

Why this answer

Increasing gp2 storage from 500 GB to 1 TB raises the baseline IOPS from 1,500 (3 IOPS/GB) to 3,000, and also increases the burst credit pool size. With WriteIOPS averaging 10,000, the volume is depleting burst credits faster than they can be replenished, causing BurstBalance to drop to 20% and latency to spike. Doubling the storage provides a higher baseline and more credits, reducing reliance on burst and resolving the latency issue without changing instance class or adding cost for reserved IOPS.

Exam trap

AWS often tests the misconception that increasing instance size (vCPUs/memory) resolves storage I/O bottlenecks, when in fact gp2 burst credit exhaustion is a storage-level issue that requires increasing volume size or switching to provisioned IOPS.

How to eliminate wrong answers

Option A is wrong because a read replica cannot offload write traffic; it only serves read queries and does not reduce write latency on the primary. Option B is wrong because upgrading to a db.r5.16xlarge increases CPU and memory but does not change the gp2 volume's IOPS baseline or burst credit behavior, so write latency caused by storage exhaustion would persist. Option C is wrong because Performance Insights and caching address query performance and read load, not the underlying gp2 burst credit depletion causing high write latency; the workload is write-heavy with frequent small transactions, not slow queries.

Practice this question →

170

MCQhard

A company is running a MongoDB-compatible Amazon DocumentDB cluster. The application is experiencing high write latency during peak hours. The cluster has one writer and two readers. The DBA suspects that the issue is due to a large number of indexes on the collection. What should the DBA do to verify this hypothesis without affecting production?

A.Take a snapshot of the cluster and restore it to a new cluster to test index changes.

B.Use AWS DMS to replicate data to a test cluster and perform index testing there.

C.Enable the profiler on the production cluster to capture slow queries and analyze the index usage.

D.Create a clone of the cluster using the 'Clone Cluster' feature and test index changes on the clone.

AnswerD

Cloning creates a copy of the cluster quickly without impacting the source.

Why this answer

Cloning the cluster using the 'Clone Cluster' feature creates a separate, isolated environment where index changes can be tested without affecting production. Option A is incorrect because taking a snapshot and restoring to a new cluster is time-consuming and does not provide a real-time copy of the current state for testing. Option B is incorrect because AWS DMS is designed for database migration, not for testing index performance.

Option C is incorrect because enabling the profiler captures slow queries but does not allow testing of index changes; it only helps identify existing issues.

Practice this question →

171

MCQeasy

A DBA sees the above error log entries for an Amazon RDS for PostgreSQL DB instance. What is the most likely cause?

A.The user 'myuser@mycompany.com' is using an incorrect password.

B.There is a network connectivity issue between the client and the database.

C.The database has reached its maximum number of connections.

D.The user 'myuser@mycompany.com' does not exist in the database.

AnswerA

Authentication failure typically indicates wrong password.

Why this answer

The error log entry shows authentication failure for user 'myuser@mycompany.com' with message 'password authentication failed for user'. In PostgreSQL, this specific error indicates the client provided an incorrect password for an existing user. The log does not show any connection limit exceeded or network timeout messages, and the user is referenced by name, confirming the user exists.

Exam trap

The trap here is that candidates confuse 'authentication failed' with 'user does not exist', but PostgreSQL deliberately returns different error messages for nonexistent roles versus incorrect passwords to avoid information leakage.

How to eliminate wrong answers

Option B is wrong because network connectivity issues would produce timeout or 'no route to host' errors, not password authentication failures. Option C is wrong because reaching maximum connections generates 'FATAL: sorry, too many clients already' errors, not password authentication failures. Option D is wrong because if the user did not exist, the error would be 'FATAL: role "myuser@mycompany.com" does not exist', not a password authentication failure.

Practice this question →

172

MCQmedium

A database engineer is troubleshooting an Amazon Aurora MySQL DB cluster that is experiencing frequent failovers. The failovers are occurring during periods of low write activity. The engineer notices that the primary instance's writer endpoint is being used for read traffic from a legacy application. What is the most likely cause of the failovers?

A.A high replication lag between the primary and replica instances

B.A long-running write transaction is blocking the failover

C.Insufficient storage for the Aurora cluster volume

D.The primary instance is overloaded due to read traffic from the writer endpoint

AnswerD

Directing read traffic to the writer endpoint increases load, leading to failovers.

Why this answer

Using the writer endpoint for read traffic can overload the primary instance, leading to resource contention and failovers even during low write activity. Option A is incorrect because replication lag is not a direct cause of failovers; failovers are typically due to primary instance issues. Option B is incorrect because a long-running write transaction would not cause failovers during low write activity; it might cause replication lag but not failover.

Option C is incorrect because insufficient storage for the Aurora cluster volume would cause storage-related errors, not failovers; Aurora automatically scales storage.

Practice this question →

173

MCQhard

A company is running Amazon Redshift and notices that queries are slow. The administrator runs the STL_ALERT_EVENT_LOG and sees many 'Nested Loop Join' alerts. What is the MOST likely cause?

A.Insufficient concurrency scaling.

B.Incorrect sort keys on the tables.

C.Insufficient compression on the tables.

D.Incorrect distribution keys on the tables.

AnswerD

Missing dist keys cause large data movement.

Why this answer

Nested loop joins in Amazon Redshift occur when the query optimizer cannot use a hash join or merge join, often because the data required for the join is not co-located on the same compute nodes. Incorrect distribution keys cause data to be distributed unevenly or in a way that forces large amounts of data to be redistributed across the network, leading to nested loop joins as a fallback. This is the most likely cause because the STL_ALERT_EVENT_LOG specifically flags nested loop joins as a performance issue related to data distribution.

Exam trap

The trap here is that candidates often confuse distribution keys with sort keys or compression, assuming that poor data organization or storage efficiency directly causes join performance issues, when in fact the root cause is data co-location and the join algorithm selected by the optimizer.

How to eliminate wrong answers

Option A is wrong because insufficient concurrency scaling affects query throughput under high concurrency, not the join method chosen by the optimizer; nested loop joins are a query plan issue, not a resource contention issue. Option B is wrong because incorrect sort keys primarily affect range-restricted scans and min/max pruning, not the join strategy; sort keys do not directly cause nested loop joins. Option C is wrong because insufficient compression reduces I/O efficiency and storage footprint but does not change the join algorithm; the optimizer still chooses join methods based on data distribution and statistics, not compression.

Practice this question →

174

Multi-Selecthard

A company uses Amazon DynamoDB for a high-traffic gaming leaderboard. The table has a partition key of 'game_id' and a sort key of 'score'. During a tournament, the application experiences throttling on a single partition. The application uses strongly consistent reads. Which TWO actions should be taken to resolve the throttling and maintain performance?

Select 2 answers

A.Redesign the partition key to include a random suffix to distribute writes across partitions.

B.Enable DynamoDB Accelerator (DAX) to cache the most frequently read items.

C.Increase the provisioned read capacity units (RCU) for the table.

D.Use eventually consistent reads for the leaderboard queries where possible.

E.Create a global secondary index (GSI) with a different partition key.

AnswersA, D

A write-sharding pattern ensures even distribution of write traffic and avoids hot partitions.

Why this answer

The correct actions are A and D. Option A: Redesigning the partition key to include a random suffix (e.g., appending a shard number) distributes write traffic across multiple partitions, preventing a single hot partition from throttling. Option D: For a leaderboard, eventually consistent reads are sufficient for non-critical queries, reducing read capacity consumption and alleviating read-side throttling.

Option B (DAX) only caches reads and does not help with write throttling. Option C (increasing RCU) addresses read capacity, not write distribution to fix the hot partition. Option E (GSI) creates a separate index but does not redistribute writes on the base table's primary partition.

Practice this question →

175

Multi-Selectmedium

A company runs an Amazon RDS for MySQL DB instance with Multi-AZ. The application team reports that a recent schema change caused a significant increase in replication lag. The instance has a read replica in a different AWS Region. Which THREE actions should be taken to reduce replication lag?

Select 3 answers

A.Identify and optimize any long-running transactions or DDL statements that may be blocking replication.

B.Increase the DB instance class for both the primary and read replica to improve performance.

C.Monitor the ReplicaLag metric in Amazon CloudWatch to understand the trend and set alarms.

D.Set the transaction isolation level to READ COMMITTED to reduce locking.

E.Disable binary logging on the primary instance to reduce I/O overhead.

AnswersA, C, D

Long-running transactions can cause replication lag by holding locks or delaying binary log events.

Why this answer

Options A, C, and D are correct actions to reduce replication lag. A: Optimizing long-running transactions or DDL statements helps prevent blocking on the primary that can cause lag. C: Monitoring the ReplicaLag metric in CloudWatch allows you to identify trends and set alarms for proactive management.

D: Setting the transaction isolation level to READ COMMITTED reduces locking on the primary, which can decrease replication lag. Option B is not the best action: increasing the instance class may improve performance but does not directly address the root cause of lag from schema changes. Option E is incorrect because disabling binary logging would stop replication entirely, not reduce lag.

Practice this question →

176

Multi-Selectmedium

A company runs an Amazon RDS for SQL Server database. The database size is 500 GB. The company needs to create a test environment with a copy of the production database that refreshes daily. Which THREE steps should be taken to automate this? (Choose THREE.)

Select 3 answers

A.Use AWS Backup or AWS Lambda to take a manual snapshot of the production DB instance daily.

B.Update the test application's connection string to point to the new test instance.

C.Create a read replica of the production instance and promote it to the test environment.

D.Restore the snapshot to a new DB instance in the test environment.

E.Enable automated backups on the test instance after restoration.

AnswersA, B, D

Creates a consistent snapshot.

Why this answer

Creates a manual snapshot of the production DB instance daily using AWS Backup or Lambda. Option D restores that snapshot to a new DB instance in the test environment. Option B updates the test application's connection string to point to the new test instance.

Option C is incorrect because a read replica is not a full copy and promoting it would disrupt replication, not provide a daily refreshed copy. Option E is unnecessary because the restored instance can have automated backups enabled, but it is not a required step for the daily refresh process.

Practice this question →

177

Multi-Selectmedium

A company is using Amazon RDS for PostgreSQL with Multi-AZ deployment. The database administrator needs to perform a minor version upgrade with minimal downtime. Which TWO actions should the administrator take? (Choose TWO.)

Select 2 answers

A.Disable Multi-AZ before the upgrade to reduce complexity.

B.Test the upgrade on a separate non-production environment first.

C.Modify the DB parameter group to set the new version parameters.

D.Schedule the upgrade during the maintenance window.

E.Take a snapshot of the DB instance before the upgrade.

AnswersB, D

Testing ensures compatibility and reduces risk of issues during the production upgrade.

Why this answer

Options B and D are correct. Testing the upgrade on a separate non-production environment first (B) is a best practice to verify compatibility and avoid issues. Scheduling the upgrade during the maintenance window (D) allows Amazon RDS to perform the upgrade automatically with a failover, minimizing downtime.

Option A is incorrect because disabling Multi-AZ would eliminate the failover benefit and cause downtime. Option C is incorrect because minor version upgrades do not require modifying the DB parameter group; AWS RDS handles parameter compatibility automatically. Option E is incorrect because taking a snapshot is a good practice for recovery but does not reduce downtime during the upgrade.

Practice this question →

178

Multi-Selecteasy

A company is using Amazon RDS for MariaDB. The database administrator wants to monitor the database for slow queries. Which TWO services can be used to achieve this? (Choose TWO.)

Select 2 answers

A.Amazon RDS Performance Insights

B.AWS Config

C.Amazon CloudWatch Logs

D.Amazon RDS Enhanced Monitoring

E.AWS CloudTrail

AnswersA, C

Performance Insights helps identify and analyze slow queries.

Why this answer

Amazon RDS Performance Insights (Option A) is correct because it provides a database performance tuning and monitoring feature that visualizes database load and allows you to identify and analyze slow queries by filtering on SQL statements, waits, and hosts. Amazon CloudWatch Logs (Option C) is correct because you can enable the slow query log for MariaDB and publish those logs to CloudWatch Logs, then use CloudWatch Logs Insights to query and filter for slow queries based on execution time.

Exam trap

The trap here is that candidates often confuse Enhanced Monitoring (OS-level metrics) with Performance Insights (database-level query analysis) or assume CloudTrail can capture database queries, when in fact CloudTrail only records AWS API calls, not SQL statements.

Practice this question →

179

MCQhard

A company's Amazon RDS for PostgreSQL instance is experiencing high CPU utilization. The DB instance is a db.r5.xlarge with 500 GB of General Purpose SSD (gp2) storage. The application team reports that write-intensive batch jobs are causing latency. The DBA notices that the ReadIOPS metric is low but WriteIOPS is consistently near the instance's maximum. Which action would most effectively reduce write latency?

A.Change the storage type to Provisioned IOPS (io1) with the same IOPS.

B.Increase the allocated storage to 1,000 GB to increase baseline IOPS.

C.Add a read replica to distribute read traffic.

D.Enable Multi-AZ deployment to offload writes to a standby instance.

AnswerB

For gp2 volumes, baseline IOPS is 3 per GB, up to 16,000 IOPS. Increasing storage size increases baseline IOPS and burst balance duration, reducing write latency.

Why this answer

Increasing the allocated storage for a gp2 volume from 500 GB to 1,000 GB doubles the baseline IOPS from 1,500 to 3,000 (gp2 baseline is 3 IOPS per GB). Since the write-intensive workload is hitting the maximum IOPS of the current volume, this increase directly raises the IOPS ceiling, reducing write latency without changing storage type.

Exam trap

The trap here is that candidates often assume Provisioned IOPS (io1) is always the answer for high IOPS workloads, but the question specifies that WriteIOPS is 'consistently near the instance's maximum' — increasing gp2 storage is a simpler, cost-effective way to raise the IOPS ceiling without changing storage type.

How to eliminate wrong answers

Option A is wrong because changing to io1 with the same IOPS (e.g., 1,500) does not increase the IOPS ceiling; it only provides consistent performance, but the bottleneck remains. Option C is wrong because a read replica offloads read traffic, not write traffic; writes still go to the primary instance, so write latency is unaffected. Option D is wrong because Multi-AZ deployment provides a standby for failover and offloads read traffic for the synchronous standby, but writes are still committed on the primary and replicated synchronously; it does not reduce write latency on the primary.

Practice this question →

180

MCQhard

A database specialist sees the above error log entries from an Amazon RDS for MySQL DB instance. Which action should be taken to resolve the issue?

A.Increase the allocated storage for the DB instance.

B.Restore the DB instance from the latest automated snapshot.

C.Reboot the DB instance to clear the error.

D.Run the mysqlcheck command to repair the corrupted tables.

AnswerB

Correct: Restoring from a clean snapshot is the standard recovery method for page corruption.

Why this answer

The error log entries indicate corruption in the InnoDB tablespace, likely due to a crash or hardware failure. Restoring from the latest automated snapshot is the correct action because it provides a consistent, point-in-time recovery of the entire database, including all table structures and data, without relying on potentially corrupted files. This approach ensures data integrity and avoids further damage that could occur from attempting repairs on a corrupted instance.

Exam trap

The trap here is that candidates often confuse table-level repair commands (like mysqlcheck or REPAIR TABLE) with the need for a full database restore, not realizing that InnoDB corruption typically requires point-in-time recovery from a backup rather than a simple repair operation.

How to eliminate wrong answers

Option A is wrong because increasing allocated storage addresses disk space issues, not data corruption; the error logs point to logical corruption, not a full filesystem. Option C is wrong because rebooting the DB instance will not repair corrupted tables; it may even cause the instance to fail to start if the corruption is severe, and it does not fix underlying data integrity problems. Option D is wrong because mysqlcheck is designed for MyISAM tables and does not repair InnoDB tables; InnoDB corruption requires recovery from a backup or using the innodb_force_recovery parameter, not a simple command-line repair tool.

Practice this question →

181

MCQhard

A database specialist runs the CLI command shown. The output indicates the instance is Multi-AZ and available. The application, however, is experiencing intermittent connection timeouts to the database endpoint. What is the most likely cause?

A.A failover occurred due to a Multi-AZ event.

B.The MySQL engine version is not supported.

C.The DB instance class is not available in the current region.

D.The database storage is full.

AnswerA

Failover causes transient connection timeouts until DNS updates.

Why this answer

A Multi-AZ failover can cause transient connection timeouts if the application does not have proper retry logic. The instance status shows 'available', so storage is not full (eliminating D). MySQL 8.0.28 is a supported version (eliminating B).

The instance class db.r5.xlarge is current generation and available in the region (eliminating C). Therefore, the intermittent timeouts are most likely due to a failover event.

Practice this question →

182

MCQhard

A company's RDS for SQL Server instance has been running for 3 years. The DBA notices that the DB instance's allocated storage (500 GB) is 80% full. The application is write-heavy. The DBA needs to increase storage without downtime. What is the most efficient way to achieve this?

A.Attach an additional EBS volume to the DB instance.

B.Create a new DB instance with larger storage and migrate data.

C.Take a snapshot and restore a new instance with larger storage.

D.Use the Modify DB Instance API to increase allocated storage.

AnswerD

RDS allows storage modification online for SQL Server.

Why this answer

RDS for SQL Server supports modifying storage online using the Modify DB Instance API without downtime. Option A is incorrect because you cannot attach additional EBS volumes directly to an RDS instance; RDS manages its own storage. Option B is incorrect because creating a new instance and migrating data would require downtime.

Option C is incorrect because taking a snapshot and restoring a new instance also incurs downtime and is less efficient than modifying storage in place.

Practice this question →

183

MCQmedium

A company runs an Amazon ElastiCache for Redis cluster as a caching layer for a high-traffic web application. The cluster has one primary and two replica nodes. Recently, the application experienced increased latency. Monitoring shows that the CPU utilization of the primary node is consistently above 80%, while replicas are below 40%. The application uses the cluster mode disabled. The operations team needs to reduce the CPU load on the primary node. Which solution should they implement?

A.Implement a read-through cache pattern using DynamoDB Accelerator (DAX).

B.Upgrade the primary node to a larger instance type.

C.Configure the application to use read replicas for read queries by using the reader endpoint.

D.Add more replica nodes to the cluster.

AnswerC

Replicas can serve read traffic, lowering primary CPU.

Why this answer

Offloading read traffic to replicas reduces CPU load on the primary node. Configuring the application to use the reader endpoint for read queries distributes read operations across replicas, lowering the primary's CPU utilization. Option A (DAX) is unrelated to ElastiCache Redis.

Option B (vertical scaling) may help but is often more expensive and does not leverage replicas. Option D (adding replicas) does not reduce primary CPU unless reads are directed to them.

Practice this question →

184

MCQmedium

Refer to the exhibit. An IAM policy is attached to a role used by an application running on Amazon EC2. The application needs to read items from a DynamoDB table named 'MyTable' and create manual snapshots of an RDS instance. What is the issue with the policy?

A.The policy does not grant any DynamoDB read actions (e.g., GetItem, Query, Scan) required for the application.

B.The policy allows 'DescribeDBInstances' on all resources, which is too permissive.

C.The policy uses 'Resource': '*' for DynamoDB, which should be the table ARN.

D.The policy grants 'CreateDBSnapshot' on all resources, which is insufficient; it should be scoped to the specific DB instance.

AnswerA

The application needs to read items, but the policy only allows PutItem and DeleteItem.

Why this answer

The policy only grants 'PutItem' and 'DeleteItem' actions on the DynamoDB table, but the application needs to read items (e.g., 'GetItem' or 'Scan'). The missing DynamoDB read actions are the primary issue. Option B is incorrect because 'DescribeDBInstances' on all resources is not inherently too permissive for this use case; the key issue is the missing read permissions.

Option C is incorrect because using 'Resource': '*' for DynamoDB is acceptable for read actions if the table ARN is not specified; it is not a policy structure problem. Option D is incorrect because 'CreateDBSnapshot' on '*' is allowed, but the policy correctly scopes RDS actions to 'Resource': '*' which includes all DB instances; the critical flaw remains the missing DynamoDB read actions.

Practice this question →

185

MCQhard

A company uses Amazon RDS for MySQL with Multi-AZ deployment. The primary instance fails and the database administrator notices that the failover did not happen automatically. What is the MOST likely cause?

A.The DB instance is in a public subnet.

B.Enhanced Monitoring is disabled.

C.The standby instance is in a different VPC.

D.The DB instance is configured as a read replica.

AnswerD

Read replicas do not have automatic failover.

Why this answer

A read replica in Amazon RDS for MySQL is a separate DB instance that asynchronously replicates from a source instance. It does not support automatic failover; if the source instance fails, the read replica remains a standalone read-only instance and does not become the primary. Multi-AZ failover only occurs between a primary and its standby in the same region, not to a read replica.

Exam trap

The trap here is that candidates confuse read replicas with Multi-AZ standby instances, assuming a read replica can automatically take over during a failure, when in fact only a Multi-AZ standby provides automatic failover.

How to eliminate wrong answers

Option A is wrong because a public subnet does not affect Multi-AZ automatic failover; failover is a DNS and network-level mechanism that works regardless of subnet type. Option B is wrong because Enhanced Monitoring is a performance metrics feature and has no role in triggering or blocking automatic failover. Option C is wrong because Multi-AZ requires the standby to be in a different Availability Zone but within the same VPC; a standby in a different VPC would not be part of the same RDS deployment and would not be configured as a Multi-AZ standby.

Practice this question →

186

Multi-Selectmedium

A company is using Amazon DynamoDB with a global table for a multi-region application. The application is deployed in us-east-1 and eu-west-1. The company notices that write latency to eu-west-1 is higher than expected. Which TWO actions can reduce write latency? (Choose TWO.)

Select 2 answers

A.Use DynamoDB Accelerator (DAX) for write caching.

B.Write data to the local Region and let global tables replicate to other Regions.

C.Use strongly consistent reads in the application.

D.Use a single-region table in eu-west-1 and replicate data manually.

E.Increase the write capacity of the table in eu-west-1.

AnswersA, B

DAX can be configured as a write-through cache, which reduces write latency by caching writes locally and flushing them asynchronously to DynamoDB.

Why this answer

DAX can be used as a write-through cache for DynamoDB, reducing write latency by caching writes locally and asynchronously flushing to the primary table. Option B is correct because writing to the local region (e.g., us-east-1) and allowing global tables replication to eu-west-1 reduces the latency for the local write operation; the replication happens asynchronously. Option C is incorrect because strongly consistent reads do not affect write operations.

Option D is incorrect because using a single-region table in eu-west-1 would eliminate the multi-region capability and manual replication adds complexity and potential latency. Option E is incorrect because increasing write capacity does not reduce latency; it increases throughput but not speed.

Exam trap

A common mistake is to think that increasing write capacity reduces write latency, but capacity is about throughput, not latency.

Practice this question →

187

MCQhard

A database administrator runs the 'describe-db-clusters' CLI command and sees the above output. The earliest restorable time is December 1, 2023. However, the backup retention period is 7 days. Why is the earliest restorable time earlier than 7 days from now (assuming today is December 5, 2023)?

A.The backup retention period is misconfigured; it should be 30 days.

B.The cluster was created on December 1, so that is the earliest point available.

C.Automated backups are only kept for 7 days, but manual snapshots extend the recovery window.

D.The 'earliestRestorableTime' is calculated based on the latest transaction log, not backup retention.

AnswerC

Manual snapshots, when taken, can extend the earliest restorable time beyond the automated backup retention period.

Why this answer

The earliest restorable time shown is December 1, 2023, which is earlier than the 7-day backup retention period (which would be November 28, 2023 from today December 5, 2023). This discrepancy occurs because manual snapshots taken prior to the automated backup window extend the restorable range. While automated backups are retained for only 7 days, manual snapshots persist independently and can provide restore points further back in time, effectively making the earliest restorable time earlier than the automated retention period alone would allow.

Practice this question →

188

Multi-Selectmedium

A company is migrating an on-premises Oracle database to Amazon RDS for Oracle. The database is 2 TB and has a high number of small transactions. The company needs to minimize downtime during the migration. Which TWO strategies should be used together? (Choose two.)

Select 2 answers

A.Use Oracle Data Pump to export the database and import into RDS.

B.Set up a VPN connection between on-premises and AWS for direct database link.

C.Create a manual snapshot of the RDS instance during the migration.

D.Use AWS Database Migration Service (DMS) to perform a full load and ongoing replication.

E.Configure change data capture (CDC) on the source database and apply to RDS.

AnswersD, E

DMS can migrate data with minimal downtime by using continuous replication after the full load.

Why this answer

AWS DMS with ongoing replication (CDC) is the correct choice because it allows a full load of the 2 TB database followed by continuous replication of changes, minimizing downtime by keeping the target RDS instance nearly synchronized with the source until cutover. This approach is specifically designed for large databases with high transaction volumes, as it reduces the final outage window to seconds or minutes.

Exam trap

The trap here is that candidates often confuse 'minimizing downtime' with 'zero downtime' and select Data Pump (A) because it is a familiar Oracle tool, failing to recognize that DMS with CDC is the only option that provides near-continuous replication and a short cutover window.

Practice this question →

189

MCQmedium

A company's production RDS for PostgreSQL instance experienced a failover to the standby replica. After the failover, the application team reports that writes are failing with 'ReadOnlyException: Connection is read-only'. The DB instance status in the RDS console shows 'available'. Which action should the database administrator take to resolve the issue?

A.Update the application connection string to use the new writer endpoint from the RDS console.

B.Create a read replica and promote it to a new primary.

C.Reboot the DB instance to force a failover back to the original primary.

D.Modify the DB instance to enable Multi-AZ again.

AnswerA

The application must connect to the writer endpoint after failover.

Why this answer

After a Multi-AZ failover, the standby becomes the primary, and the old primary becomes a read-only standby. The application must update its connection string to use the new writer endpoint (which automatically points to the new primary) to perform writes. Option B is incorrect because creating a read replica and promoting it is unnecessary; Multi-AZ already provides a standby.

Option C is incorrect because rebooting would force another failover but does not change the endpoint. Option D is incorrect because Multi-AZ is already enabled; modifying it again does not resolve the issue.

Practice this question →

190

MCQeasy

A company is using Amazon DynamoDB with on-demand capacity mode. The company notices that during a flash sale, the write requests are throttled. What is the most likely cause of the throttling?

A.The table has reached its maximum throughput limit for on-demand mode.

B.The table is not configured for auto scaling.

C.A hot partition is causing throttling even though the table uses on-demand capacity.

D.The provisioned write capacity units (WCUs) are set too low.

AnswerC

On-demand tables can still throttle if a single partition receives more writes than its limit (1000 writes/sec).

Why this answer

For a DynamoDB table in on-demand capacity mode, throttling is typically caused by a hot partition. On-demand mode automatically scales table-level throughput, but if a single partition key is heavily accessed (e.g., during a flash sale), that partition can exceed its individual capacity limit, resulting in throttled write requests. This is the most likely cause, making option C correct.

Option A (maximum throughput limit) does not apply because on-demand has no fixed maximum; option B (auto scaling) is irrelevant since on-demand manages scaling automatically; option D (provisioned WCUs) is not used in on-demand mode.

Practice this question →

191

MCQmedium

A company manages multiple RDS for PostgreSQL instances in different AWS accounts. The security team requires that all DB instances have automated backups enabled with a retention period of at least 14 days. How can the company enforce this policy across all accounts?

A.Use AWS CloudTrail to monitor backup settings and alert on non-compliance.

B.Apply a service control policy (SCP) that requires backup retention.

C.Create an IAM policy that denies creating DB instances without backup retention.

D.Use AWS Config rules to evaluate and auto-remediate non-compliant instances.

AnswerD

AWS Config rules can evaluate RDS instances for backup retention and trigger auto-remediation to fix non-compliant resources.

Why this answer

AWS Config with managed rules can evaluate RDS instances and enforce backup settings. Option A is wrong because CloudTrail only logs API calls for auditing, it does not enforce configuration. Option B is wrong because Service Control Policies (SCPs) restrict permissions at the organization level but cannot directly require backup retention on resources.

Option C is wrong because IAM policies can control permissions to create instances but cannot enforce backup retention settings on existing instances; they can deny creation if certain conditions are not met, but that is not as comprehensive as AWS Config auto-remediation.

Practice this question →

192

Multi-Selecteasy

A company is designing a disaster recovery strategy for an Amazon Aurora MySQL database with a recovery point objective (RPO) of 1 second and a recovery time objective (RTO) of 1 minute. Which TWO solutions meet these requirements? (Choose TWO.)

Select 2 answers

A.Use a Multi-AZ DB cluster deployment.

B.Take manual snapshots every hour and copy to another region.

C.Enable automated backups and perform point-in-time recovery in another region.

D.Deploy an Aurora Global Database with a secondary region.

E.Configure cross-Region read replicas and promote in disaster.

AnswersD, E

Global Database offers low RPO/RTO.

Why this answer

(Aurora Global Database) provides cross-region replication with an RPO of typically less than 1 second and RTO of approximately 1 minute, meeting the requirements. Option E (cross-Region read replicas) can be promoted to a standalone primary in under a minute, and replication lag is sub-second, achieving RPO of 1 second. Option A (Multi-AZ DB cluster) is confined to a single region and does not provide DR across regions.

Option B (manual snapshots every hour) results in an RPO of up to 1 hour, far exceeding the 1-second requirement. Option C (automated backups) has a default backup interval of 5 minutes, resulting in an RPO of at least 5 minutes, and point-in-time recovery typically takes longer than 1 minute.

Practice this question →

193

MCQhard

A company uses Amazon DynamoDB with on-demand capacity for a gaming leaderboard. During a promotional event, write traffic spikes 10x, causing occasional 'ProvisionedThroughputExceededException' errors. The application retries with exponential backoff, but latency increases. The team notices that the 'ThrottledWriteRequests' metric spikes. What is the MOST cost-effective solution to handle these unpredictable spikes?

A.Continue using on-demand capacity but ensure the table has no throttling.

B.Implement DynamoDB Accelerator (DAX) to reduce read load.

C.Switch to provisioned capacity with auto scaling configured for the expected peak.

D.Use an Amazon SQS queue to buffer write requests before DynamoDB.

AnswerC

Correct. Switching to provisioned capacity with auto scaling allows the table to scale based on actual traffic patterns, handling spikes cost-effectively without manual intervention. This directly addresses the throughput exception.

Why this answer

The scenario describes a DynamoDB table using on-demand capacity but experiencing 'ProvisionedThroughputExceededException', which is an error specific to provisioned capacity. This inconsistency suggests the table is actually using provisioned capacity. Therefore, the best solution is to switch to provisioned capacity with auto scaling properly configured to handle the peak traffic, ensuring cost-effectiveness by scaling only when needed.

Option A is invalid because on-demand tables do not throw this error. Option B (DAX) addresses read latency, not write throttling. Option D (SQS) adds complexity and latency and is not the most cost-effective for unpredictable write spikes.

Exam trap

The question contains an inconsistency: on-demand capacity is stated, but the error 'ProvisionedThroughputExceededException' is only relevant to provisioned capacity. Recognize that the best solution addresses the underlying provisioning model.

Practice this question →

194

MCQmedium

An e-learning platform uses Amazon Aurora MySQL for its database. The application runs reporting queries that scan large portions of the database, causing high CPU utilization on the primary instance. The primary instance is a db.r5.2xlarge with 64 GB memory. The reporting queries are not time-sensitive but need to return results within 5 minutes. The operations team wants to reduce the impact on the primary instance without increasing costs significantly. Which action should be taken?

A.Modify the DB cluster parameter group to enable result set caching

B.Create an Aurora Replica and configure the reporting application to connect to the replica endpoint

C.Use Amazon ElastiCache to cache the reporting results

D.Increase the primary instance to db.r5.4xlarge

AnswerB

Creating an Aurora Replica allows reporting queries to run on a separate read-only instance, reducing CPU on the primary. It is cost-effective because you only pay for the replica instance.

Why this answer

To create an Aurora Replica and direct reporting traffic to it. This offloads the heavy reporting queries from the primary instance, reducing CPU impact without significant cost increase. Aurora Replicas share the same storage and are cost-effective for read-heavy workloads.

Practice this question →

195

MCQmedium

A company is running an Amazon RDS for SQL Server DB instance. The database administrator needs to perform a major version upgrade. What is the recommended approach to minimize downtime?

A.Use AWS Database Migration Service (DMS) to migrate the database to a new instance with the new version.

B.Create a Read Replica of the DB instance with the new version, promote it to a standalone instance, and then redirect application traffic.

C.Take a snapshot of the DB instance and restore it with the new version.

D.Modify the DB instance directly and apply the new version during the maintenance window.

AnswerA

Correct. AWS DMS can migrate to a new RDS instance with a new major version while minimizing downtime through continuous replication and a cutover.

Why this answer

For Amazon RDS for SQL Server, creating a Read Replica with a different major version is not supported. Therefore, the recommended approach to minimize downtime is to use AWS Database Migration Service (DMS) to migrate the database to a new instance with the new version. DMS allows for minimal downtime by continuously replicating data and then switching over.

Options C and D cause significant downtime during the snapshot/restore or direct upgrade process.

Exam trap

Candidates may assume that Read Replicas can be used for major version upgrades on all RDS engines, but SQL Server does not support cross-version Read Replicas.

Practice this question →

196

MCQmedium

An IAM user has the policy shown. The user is trying to restore a DB instance from a manual snapshot using the AWS CLI. The restore fails with an access denied error. What is the most likely reason?

A.The policy does not grant the rds:RestoreDBInstanceFromDBSnapshot action on the DB instance resource.

B.The rds:RestoreDBInstanceFromDBSnapshot action is misspelled.

C.The snapshot is encrypted and the user does not have permission to use the KMS key.

D.The snapshot resource ARN does not include the specific snapshot ID.

AnswerA

The restore action requires permission on both the snapshot and the DB instance.

Why this answer

The restore operation (rds:RestoreDBInstanceFromDBSnapshot) requires permission on both the snapshot resource and the DB instance resource. The IAM policy grants the action on snapshot resources (as indicated by the snapshot resource ARN), but does not grant it on the DB instance resource (the DB instance resource ARN is missing or not included for the restore action). Therefore, the restore fails with access denied even though the user can list snapshots.

Option A correctly identifies this issue. Option B is incorrect because the action is spelled correctly (the policy allows the action on snapshots). Option C is incorrect because there is no mention of encryption or KMS key in the scenario.

Option D is incorrect because the snapshot resource ARN can use a wildcard; the lack of a specific snapshot ID is not the issue.

Practice this question →

197

Multi-Selectmedium

A company is migrating its on-premises Oracle database to Amazon RDS for Oracle. The database is 2 TB in size and the network bandwidth is 100 Mbps. The migration must have minimal downtime. Which TWO approaches should be used together to achieve this?

Select 2 answers

A.Take a full backup of the source database and restore it to Amazon RDS.

B.Increase the network bandwidth to 1 Gbps to speed up the transfer.

C.Use AWS Schema Conversion Tool (SCT) to convert the schema before migration.

D.Use AWS Database Migration Service (DMS) with ongoing replication to keep the target in sync.

E.Use AWS Snowball to transfer the initial data load to Amazon S3, then use DMS to migrate the remaining changes.

AnswersD, E

AWS DMS with ongoing replication allows continuous data replication with minimal downtime, making it suitable for this requirement.

Why this answer

AWS DMS with ongoing replication allows continuous data replication with minimal downtime. AWS Snowball can be used for the initial large data transfer to avoid prolonged network transfer. Option A (taking a full backup and restoring) would cause downtime; Option B (increasing bandwidth) is not feasible quickly; Option C (SCT) is for schema conversion, not data migration; Option D (DMS with ongoing replication) is correct; Option E (Snowball for initial load then DMS for ongoing changes) minimizes network transfer time.

Practice this question →

198

MCQhard

A company is migrating an on-premises Oracle database to Amazon Aurora PostgreSQL. The database has several stored procedures that use Oracle-specific PL/SQL. The team needs to minimize manual code changes. Which AWS service or tool should be used to automate the conversion of the stored procedures?

A.AWS Schema Conversion Tool (AWS SCT)

B.Amazon Aurora PostgreSQL native compatibility

C.AWS Database Migration Service (AWS DMS)

D.Manual rewrite using PostgreSQL PL/pgSQL

AnswerA

AWS SCT automates the conversion of database schemas and code, including PL/SQL, to target database engines.

Why this answer

The AWS Schema Conversion Tool (AWS SCT) is the correct choice because it automates the conversion of Oracle PL/SQL stored procedures to PostgreSQL PL/pgSQL, minimizing manual code changes. Option B (Amazon Aurora PostgreSQL native compatibility) does not automatically convert Oracle PL/SQL; it only supports some PostgreSQL features. Option C (AWS DMS) handles data migration, not schema or code conversion.

Option D (manual rewrite) is not automated and contradicts the requirement to minimize manual changes.

Practice this question →

199

MCQhard

A database administrator is using AWS DMS to migrate an on-premises Oracle database to Amazon RDS for PostgreSQL. The migration has been running for several hours, but the full load phase is taking much longer than expected. The CPU utilization on the DMS replication instance is consistently below 10%. What is the MOST likely cause of the slow performance?

A.The source database has large LOBs that are being transferred in full LOB mode.

B.The target database engine is not compatible with DMS.

C.The DMS task is not configured to use parallel tables.

D.The target RDS instance is throttling write operations due to low IOPS.

AnswerD

Low IOPS on the target can cause DMS to wait, leading to low CPU on the replication instance.

Why this answer

When the target RDS instance has insufficient IOPS, write operations are throttled, causing a bottleneck that slows down the full load phase. Even though the DMS replication instance's CPU is low (below 10%), the target database cannot keep up with the incoming data, leading to increased latency and reduced throughput. This is a common performance issue in DMS migrations where the target's provisioned IOPS are exhausted, especially during large data loads.

Exam trap

The trap here is that candidates often assume slow migration is due to DMS configuration (like LOB mode or parallelism) or source-side issues, but the low CPU on the replication instance is a key indicator that the bottleneck is on the target side, specifically I/O throttling.

How to eliminate wrong answers

Option A is wrong because large LOBs transferred in full LOB mode can slow down migration, but this would typically cause high CPU or memory usage on the DMS instance, not consistently low CPU. Option B is wrong because Amazon RDS for PostgreSQL is a fully compatible target for AWS DMS; DMS supports Oracle-to-PostgreSQL migrations natively. Option C is wrong because not using parallel tables can reduce throughput, but the DMS task would still utilize the replication instance's resources more heavily; the low CPU indicates the bottleneck is elsewhere, not in parallelism.

Practice this question →

200

MCQeasy

A company uses Amazon RDS for PostgreSQL with a single-AZ deployment. The operations team needs to ensure that the database is available during a planned maintenance event that requires a reboot. The maintenance window is set for 30 minutes. The database size is 200 GB and the application can tolerate a few minutes of downtime. Which action should the team take to minimize downtime during the reboot?

A.Take a manual snapshot before the maintenance window and restore it if needed

B.Create a read replica and promote it to primary after the reboot

C.Schedule the reboot during a low-traffic period

D.Modify the DB instance to be Multi-AZ

AnswerD

Multi-AZ provides automatic failover with minimal downtime during maintenance.

Why this answer

Enabling Multi-AZ on the RDS instance creates a standby in a different Availability Zone. During planned maintenance, Amazon RDS automatically performs a failover to the standby, resulting in minimal downtime (typically under a minute), which meets the requirement of a few minutes of downtime. Option A is incorrect because restoring from a manual snapshot takes much longer than a few minutes and is not an automated failover.

Option B is incorrect because promoting a read replica requires manual intervention and the replica may not be fully synchronized, leading to potential data loss and longer downtime. Option C merely schedules the reboot during low traffic but does not reduce the downtime of the reboot itself; the instance still becomes unavailable during the reboot.

Practice this question →

201

Multi-Selectmedium

A company is running an Amazon RDS for MySQL instance with Multi-AZ. The primary instance in us-east-1a fails, and the standby in us-east-1b is promoted. The application cannot connect after failover. Which TWO steps should the database administrator take to restore connectivity?

Select 2 answers

A.Update the application's connection string to point to the new writer endpoint.

B.Reboot the new primary instance to reset connections.

C.Update the security group to allow inbound traffic from the application.

D.Wait for DNS propagation and flush the application's DNS cache.

E.Create a read replica and promote it to a new primary.

AnswersA, D

Explicitly updating the endpoint ensures immediate connectivity.

Why this answer

After failover, the application may be connecting to the old primary instance endpoint. Updating the connection string to the DNS writer endpoint (which automatically points to the new primary) ensures connectivity. Option D is correct because after failover, the writer endpoint's DNS record updates to the new primary's IP.

Waiting for propagation and flushing the DNS cache ensures the application resolves the updated IP. Option B is incorrect because rebooting is not required and would cause further disruption. Option C is incorrect because the security group already permits traffic from the application; the issue is DNS resolution, not network access.

Option E is incorrect because creating a read replica does not solve the connectivity issue and is unnecessary for a Multi-AZ failover.

Practice this question →

202

MCQmedium

A company is running an Amazon RDS for SQL Server Multi-AZ DB instance. During a recent failover test, the application experienced a timeout of 60 seconds. The application uses a connection string that points to the CNAME of the DB instance. Which configuration change would reduce the failover time?

A.Create a read replica and promote it to a standalone instance during failover.

B.Modify the application to use the IP address of the DB instance instead of the CNAME.

C.Enable TDS Keep-Alive on the client side to detect and recover from connection drops faster.

D.Disable Multi-AZ and use a single-AZ instance to avoid failover overhead.

AnswerC

TDS Keep-Alive helps the client detect a broken connection sooner and reconnect to the new primary, reducing perceived downtime.

Why this answer

Enable TDS Keep-Alive on the client side. This allows the client to detect connection drops faster and reconnect to the new primary during failover, reducing the timeout experienced. Option A (Create a read replica and promote it) is a manual process and does not reduce failover time; it is used for read scaling or disaster recovery.

Option B (Using the IP address) is not recommended because the IP changes after failover; the CNAME automatically updates to point to the new primary, so using IP would cause longer outages. Option D (Disable Multi-AZ) would eliminate automatic failover, increasing downtime during failures.

Exam trap

Candidates may think that modifying the connection string to use IP address or disabling Multi-AZ would reduce failover time, but these are incorrect because they either prevent automatic failover or cause connection failures.

Practice this question →

203

MCQeasy

A company needs to ensure that all changes to an Amazon RDS DB instance are logged for auditing purposes. Which AWS service should be enabled?

A.AWS CloudTrail

B.AWS Config

C.Amazon GuardDuty

D.Amazon Inspector

AnswerA

CloudTrail records all RDS API calls for auditing.

Why this answer

AWS CloudTrail is the correct service because it records API activity for Amazon RDS, including calls to create, modify, or delete DB instances, as well as changes to security groups, parameter groups, and automated backups. CloudTrail logs these events to an S3 bucket or CloudWatch Logs, providing a durable audit trail for all management-plane operations on the RDS DB instance. This directly meets the requirement to log all changes for auditing purposes.

Exam trap

The trap here is that candidates often confuse AWS Config (which tracks configuration changes) with CloudTrail (which tracks API calls), but Config only records the state of resources after a change, not the who, what, or when of the API action that caused it.

How to eliminate wrong answers

Option B (AWS Config) is wrong because it evaluates resource configurations against desired policies and tracks configuration changes over time, but it does not log API-level actions or provide an audit trail of who made the change and when; it focuses on compliance and configuration drift, not operational auditing. Option C (Amazon GuardDuty) is wrong because it is a threat detection service that monitors for malicious activity using VPC Flow Logs, DNS logs, and CloudTrail events, but it does not itself log changes to RDS instances; it consumes logs for security analysis. Option D (Amazon Inspector) is wrong because it is a vulnerability assessment service that scans EC2 instances and container images for software vulnerabilities and unintended network exposure; it has no capability to log or audit changes to RDS DB instances.

Practice this question →

204

Multi-Selecthard

A company is running a production Amazon DynamoDB table with on-demand capacity. The table experiences occasional throttling during traffic spikes. The table's partition key is a timestamp, and the workload is write-heavy. The operations team needs to reduce throttling. Which THREE actions should the team take? (Choose three.)

Select 3 answers

A.Add a random suffix to the partition key to distribute writes evenly.

B.Increase the read capacity units for the table.

C.Use DynamoDB Accelerator (DAX) to cache read-heavy queries.

D.Switch to provisioned capacity mode with auto scaling.

E.Enable DynamoDB global tables to distribute write traffic.

AnswersA, C, D

This prevents hot partitions.

Why this answer

Adding a random suffix to the timestamp partition key breaks the sequential write pattern, distributing writes evenly across all partitions. This prevents hot partitions, which are the root cause of throttling in a write-heavy workload with a monotonically increasing partition key.

Exam trap

The trap here is that candidates assume on-demand capacity eliminates all throttling, but they overlook that throttling can still occur at the partition level due to uneven access patterns, which requires application-level key design changes to resolve.

Practice this question →

205

MCQhard

A company is running an Amazon DynamoDB table with on-demand capacity mode. The table experiences occasional throttling during peak hours. The application team wants to understand the read/write patterns to optimize the table design. Which approach should the database specialist take to analyze the throttling events?

A.Enable DynamoDB Accelerator (DAX) to cache reads and reduce throttling.

B.Use AWS CloudTrail to log all DynamoDB API calls and analyze the logs.

C.Switch to provisioned capacity mode with auto scaling to handle the spikes.

D.Enable CloudWatch Contributor Insights for DynamoDB to identify throttled requests.

AnswerD

Contributor Insights analyzes throttled requests and helps identify the top contributors, such as specific partition keys.

Why this answer

CloudWatch Contributor Insights analyzes high-cardinality attributes and provides detailed information about throttling requests, such as which items or partitions are causing throttling.

Practice this question →

206

MCQeasy

A developer accidentally deleted a critical table from an Amazon RDS for MySQL DB instance. Automated backups are enabled with a retention period of 7 days. The deletion occurred 3 hours ago. Which action can restore the table with minimal data loss?

A.Perform a point-in-time restore to a time just before the deletion.

B.Use AWS Database Migration Service to replicate the table from another source.

C.Restore the DB instance from the latest automated snapshot.

D.Use the MySQL binary log to replay transactions up to the deletion.

AnswerA

Point-in-time recovery allows restoring to any second within the retention period.

Why this answer

Amazon RDS point-in-time recovery (PITR) allows restoring a DB instance to any second within the backup retention period, enabling recovery to just before the table deletion, minimizing data loss. Option B is incorrect because AWS DMS is a migration tool, not a recovery mechanism for deleted data. Option C is incorrect because restoring from the latest automated snapshot would lose all changes made after that snapshot, including the 3 hours of data before deletion.

Option D is incorrect because while MySQL binary logs can be used for point-in-time recovery, RDS manages this through its automated PITR feature; manually replaying binary logs is not a supported operation in RDS.

Practice this question →

207

MCQhard

A company runs a critical e-commerce application on Amazon RDS for PostgreSQL with a db.r5.2xlarge instance and 500 GB of gp2 storage. The application experiences periodic write spikes during flash sales. During these events, the WriteIOPS metric exceeds the provisioned baseline IOPS of 1,500, and the database becomes unresponsive for several seconds. The DBA has configured a CloudWatch alarm on WriteIOPS, but the alarm triggers after the performance issue occurs. The company needs to ensure that the database can handle these spikes without downtime. The budget allows for moderate cost increases. What should the DBA do?

A.Increase the allocated storage to 1,000 GB to increase baseline IOPS to 3,000.

B.Modify the DB instance to use gp3 storage with provisioned IOPS of 5,000.

C.Enable Performance Insights to identify the problematic queries and tune them.

D.Add a read replica and redirect read traffic to it to reduce write contention.

AnswerB

gp3 provides a baseline of 3,000 IOPS for any storage size and allows provisioning additional IOPS up to 16,000, independent of storage. This handles write spikes cost-effectively.

Why this answer

Gp3 storage provides a baseline of 3,000 IOPS regardless of storage size, and allows provisioning up to 16,000 IOPS independently. By setting provisioned IOPS to 5,000, the database can handle write spikes without exceeding the IOPS limit, preventing unresponsiveness. This solution fits the moderate cost increase budget, as gp3 is typically 20% cheaper than gp2 for equivalent performance.

Exam trap

The trap here is that candidates assume increasing gp2 storage (Option A) is the only way to raise baseline IOPS, overlooking gp3's ability to provision higher IOPS independently without massive storage growth.

How to eliminate wrong answers

Option A is wrong because increasing gp2 storage to 1,000 GB only raises baseline IOPS to 3,000 (3 IOPS per GB), which may still be insufficient for severe write spikes and incurs higher storage costs without addressing burst balance depletion. Option C is wrong because Performance Insights identifies query performance issues but does not resolve IOPS bottlenecks; tuning queries cannot increase the underlying storage IOPS limit. Option D is wrong because read replicas handle read traffic only and do not reduce write IOPS on the primary instance; write spikes still affect the primary database.

Practice this question →

208

MCQhard

A company has an Amazon RDS for MySQL DB instance with automated backups enabled. The database is 500 GB in size. The company wants to create a new test database from the current state with minimal impact on production. Which approach meets these requirements?

A.Take a snapshot of the DB instance and restore it to a new instance.

B.Create a read replica and promote it to a standalone instance.

C.Use mysqldump to export the database and import into a new instance.

D.Create a manual DB snapshot from the automated backup and restore.

AnswerA

Snapshots are taken asynchronously with minimal performance impact, and restore creates a new instance.

Why this answer

Taking a snapshot of the RDS DB instance and restoring it to a new instance creates an independent copy of the database with minimal performance impact on the source. Snapshots are asynchronous and capture a consistent state without blocking writes, making this the most efficient method for creating a test database from the current production state.

Exam trap

The trap here is that candidates may confuse the ability to create a manual snapshot from an automated backup (which is not supported) with the valid option of taking a new manual snapshot directly from the DB instance, leading them to choose option D incorrectly.

How to eliminate wrong answers

Option B is wrong because promoting a read replica to a standalone instance requires the replica to first catch up with the source, which introduces replication lag and potential disruption; additionally, the read replica consumes resources on the source instance and is not designed for creating a point-in-time copy without affecting production. Option C is wrong because using mysqldump to export a 500 GB database would cause significant I/O and CPU load on the production instance, potentially impacting performance and requiring a long export time. Option D is wrong because creating a manual DB snapshot from an automated backup is not directly supported; automated backups are used for point-in-time recovery, but you cannot create a manual snapshot from an automated backup — you must either take a new manual snapshot or restore from an automated backup to a new instance.

Practice this question →

209

MCQeasy

A company needs to audit all SQL queries executed on an Amazon RDS for SQL Server database. Which AWS service should be used?

A.Amazon RDS Database Activity Streams

B.Amazon VPC Flow Logs

C.Amazon CloudWatch Logs

D.AWS CloudTrail

AnswerA

Database Activity Streams capture database activity such as SQL queries in near real-time.

Why this answer

Amazon RDS Database Activity Streams provides a near real-time feed of database activity, including SQL queries, making it ideal for auditing all SQL queries on an RDS for SQL Server database. AWS CloudTrail (Option D) logs API calls to AWS services, not SQL queries. VPC Flow Logs (Option B) capture network traffic metadata, not database queries.

Amazon CloudWatch Logs (Option C) can store and monitor logs but requires a log source; RDS does not natively send SQL logs to CloudWatch unless integrated with Database Activity Streams, which is the direct solution. Therefore, Option A is correct.

Practice this question →

210

MCQeasy

A database specialist sees the above error in the application logs for an Amazon RDS for MySQL DB instance. The application is a web server running on an EC2 instance. What is the most likely cause?

A.The database user password has expired.

B.A user attempted to execute an invalid SQL query.

C.The database connection timed out due to inactivity or network connectivity issues.

D.The database instance has run out of disk space.

AnswerC

Common cause of 'server has gone away'.

Why this answer

The error message in the application logs indicates a connection timeout, which occurs when the client (EC2 web server) cannot establish or maintain a TCP connection to the RDS MySQL DB instance within the configured timeout period. This is most commonly caused by network connectivity issues (e.g., security group rules, NACLs, route tables, or VPC misconfigurations) or prolonged inactivity that triggers the MySQL `wait_timeout` or `interactive_timeout` setting, causing the server to close the connection. Option C directly addresses this scenario, while the other options produce different error codes or symptoms.

Exam trap

The trap here is that candidates often confuse connection timeout errors with authentication or query syntax errors, but the specific error message in the logs (e.g., 'Connection timed out' or 'Lost connection to MySQL server during query') directly points to network or timeout issues, not SQL or credential problems.

How to eliminate wrong answers

Option A is wrong because an expired database user password would result in an authentication failure error (e.g., 'Access denied for user') at connection time, not a timeout error. Option B is wrong because an invalid SQL query would produce a MySQL syntax error (e.g., 'You have an error in your SQL syntax') after the connection is successfully established, not a timeout before or during the query execution. Option D is wrong because running out of disk space on the RDS instance would cause write failures or the instance to become read-only (e.g., 'The MySQL server is running with the --read-only option'), not a connection timeout.

Practice this question →

211

MCQhard

A company runs a critical application on Amazon RDS for PostgreSQL with Multi-AZ. The database has a large table (over 500 GB) that is frequently updated. The operations team notices that the primary instance's CPU usage is consistently above 90%, and the replica lag between the primary and standby is increasing during peak hours. The application can tolerate a few seconds of downtime. The team needs to reduce CPU load and improve write performance without changing the application code. Which action should be taken?

A.Upgrade the DB instance class to a larger size with more vCPUs.

B.Increase the backup retention period to reduce I/O during backups.

C.Modify the DB instance to use asynchronous replication instead of synchronous.

D.Create a read replica in the same region and offload read queries to it.

AnswerA

More CPU capacity reduces the load on the primary and helps keep up with replication.

Why this answer

Upgrading to a larger DB instance class provides additional vCPUs, reducing CPU utilization and alleviating replication lag. Option B is wrong: increasing the backup retention period does not reduce I/O during backups; it only keeps backups longer and does not address CPU or write performance. Option C is wrong: Multi-AZ RDS for PostgreSQL uses synchronous replication; switching to asynchronous replication is not possible without compromising durability, and it would not reduce CPU load on the primary.

Option D is wrong: creating a read replica offloads read queries, but it does not reduce write load or CPU usage from writes on the primary instance; replication lag may persist or increase.

Practice this question →

212

MCQhard

A company uses Amazon DynamoDB with on-demand capacity mode for a critical application. During a marketing campaign, the application experienced throttled requests despite the on-demand mode. The table has a single partition key. The database specialist notices that the throttling occurs sporadically even though overall traffic is within limits. What is the most likely cause?

A.The table's provisioned capacity is set too low.

B.The table's partition key is causing a hot partition, leading to throttling on that partition.

C.The table's auto scaling is not configured correctly.

D.The application is exceeding the DynamoDB account-level throughput limits.

AnswerB

A single partition key can cause hot partitions; DynamoDB's on-demand mode partitions data, but a single hot key can still throttle.

Why this answer

On-demand mode accommodates traffic spikes, but if a single partition key is used, all traffic goes to one partition. DynamoDB partitions data by partition key; a single hot key can throttle requests even if overall throughput is within limits. Provisioned capacity is not used.

Auto scaling is not relevant. The partition limit is a hard limit.

Practice this question →

213

MCQeasy

A company has an Amazon Redshift cluster with two dc2.large nodes. The cluster is used for daily ETL jobs and reporting. The operations team receives an alert that the cluster's disk space is 90% full. The ETL jobs are failing with 'disk full' errors. The team needs to resolve the issue quickly with minimal downtime. Which action should be taken?

A.Perform a deep copy to re-sort and reclaim space.

B.Run the VACUUM command to reclaim space from deleted rows.

C.Resize the cluster to a larger node type, such as dc2.large to ds2.xlarge, or add more nodes.

D.Unload old data to Amazon S3 and delete from the cluster.

AnswerC

Resizing increases the total storage capacity, resolving the disk full issue.

Why this answer

Resizing the cluster to a larger node type or adding more nodes increases the total storage capacity, directly addressing the disk full issue. Option A is wrong because a deep copy reorganizes data but does not increase storage capacity. Option B is wrong because VACUUM reclaims space from deleted rows but may not free enough space when the cluster is already 90% full.

Option D is wrong because unloading data to S3 removes data from the cluster only if deleted, but the statement does not include deletion; even if deleted, the space is not reclaimed until VACUUM.

Practice this question →

214

MCQmedium

A company is running a production Amazon RDS for MySQL Multi-AZ DB instance. The database experiences intermittent high latency and the CloudWatch 'ReadLatency' metric spikes during periods of heavy read traffic. The application uses a single database endpoint. What is the MOST effective way to reduce read latency without changing the application code?

A.Enable a Multi-AZ deployment with one or more readable standby replicas.

B.Implement database sharding across multiple RDS instances.

C.Enable Multi-AZ on the existing DB instance to provide a standby for failover.

D.Increase the DB instance class to a larger size.

AnswerA

Readable standby replicas in Multi-AZ allow read traffic to be directed to the standby, reducing load on the primary and lowering read latency, without code changes.

Why this answer

Enabling a Multi-AZ deployment with one or more readable standby replicas allows you to offload read traffic to the standby instances using the Read Replica endpoint, reducing load on the primary and lowering ReadLatency. Since the application uses a single database endpoint, you can use Amazon RDS's built-in reader endpoint (for a Multi-AZ DB cluster) or configure a custom DNS to distribute reads, without modifying application code. This directly addresses the intermittent high latency during heavy read traffic by scaling read capacity horizontally.

Exam trap

The trap here is that candidates confuse 'Multi-AZ' with 'readable standby' — classic Multi-AZ provides high availability but no read scaling, while the newer Multi-AZ DB cluster (or adding Read Replicas) is required to reduce read latency.

How to eliminate wrong answers

Option B is wrong because database sharding requires application code changes to route queries to the correct shard, which violates the constraint of not changing application code. Option C is wrong because enabling Multi-AZ on the existing DB instance only provides a standby for failover (not readable), which does not reduce read latency during normal operations. Option D is wrong because increasing the DB instance class to a larger size may improve performance but is less cost-effective and does not scale read capacity as efficiently as adding readable replicas, especially under intermittent heavy read traffic.

Practice this question →

215

MCQmedium

A company is using Amazon DynamoDB with on-demand capacity. The operations team notices that the number of throttled write requests has increased. Which metric should be monitored to determine if the table's write capacity is being exceeded?

A.ThrottledWriteRequests

B.WriteThrottleEvents

C.ProvisionedWriteCapacityUnits

D.ConsumedWriteCapacityUnits

AnswerB

This metric directly shows the number of throttled write requests.

Why this answer

'WriteThrottleEvents' is the Amazon CloudWatch metric that directly indicates throttled write requests. Option A is wrong because 'ThrottledWriteRequests' is not a standard CloudWatch metric name (the correct name is WriteThrottleEvents). Option C is wrong because 'ProvisionedWriteCapacityUnits' is not applicable for on-demand capacity.

Option D is wrong because 'ConsumedWriteCapacityUnits' shows actual usage, not throttling.

Practice this question →

216

MCQhard

A database administrator runs the above AWS CLI command to troubleshoot replication issues. The DB instance 'mydb' is a read replica of 'my-source-db'. The administrator notices that the replica lag is increasing. Which of the following is the MOST likely cause?

A.The source DB instance is running a different MySQL version.

B.The read replica has Multi-AZ disabled.

C.The read replica is using a smaller instance class than the source.

D.The read replica is in a different AWS Region than the source.

AnswerD

Cross-Region replication introduces network latency, causing lag.

Why this answer

When a read replica is in a different AWS Region than the source, the replication traffic must traverse the public internet or a VPN connection, introducing network latency and potential bandwidth constraints. This cross-region lag is a common cause of increasing replica lag, as the asynchronous MySQL replication relies on a single I/O thread to download the binary log events from the source, and any network delay directly impacts the replica's ability to keep up.

Exam trap

The trap here is that candidates often assume instance size (Option C) is the primary cause of replica lag, but the question explicitly mentions a cross-Region scenario (implied by the AWS CLI command targeting a different Region), making network latency the most likely culprit over compute capacity.

How to eliminate wrong answers

Option A is wrong because MySQL cross-version replication is supported as long as the source version is lower than or equal to the replica version, and version mismatch typically causes replication to fail entirely rather than just increasing lag. Option B is wrong because Multi-AZ on a read replica affects high availability and failover behavior, not the replication lag between the source and the replica. Option C is wrong because while a smaller instance class can contribute to lag if the replica lacks sufficient CPU or memory to apply changes, the most likely cause given the scenario of a cross-region replica is the network latency inherent in the geographic distance, not the instance size.

Practice this question →

217

MCQeasy

A company has an Amazon DynamoDB table with provisioned capacity. The table experiences occasional spikes in write traffic that exceed the provisioned write capacity units (WCU). Which feature should the database specialist enable to handle these spikes without throttling?

A.Enable DynamoDB burst capacity.

B.Configure DynamoDB Auto Scaling for write capacity.

C.Use DynamoDB Accelerator (DAX) to cache writes.

D.Switch to on-demand capacity mode.

AnswerB

Auto Scaling adjusts capacity automatically to handle spikes.

Why this answer

DynamoDB Auto Scaling automatically adjusts the provisioned read and write capacity based on actual traffic patterns, enabling the table to handle spikes without throttling while maintaining cost efficiency. Option A is incorrect because burst capacity provides a limited buffer for short-term spikes but can be exhausted, leading to throttling. Option C is incorrect because DAX is an in-memory cache for read operations, not writes.

Option D is incorrect because switching to on-demand capacity mode can handle spikes but may result in higher costs compared to Auto Scaling with provisioned capacity.

Practice this question →

218

MCQmedium

A company's Amazon RDS for Oracle instance is running out of storage space. The database administrator wants to add storage without downtime. Which action should be taken?

A.Delete old archived redo logs to free up space.

B.Take a snapshot and restore to a larger instance.

C.Use the 'Modify DB Instance' action to increase the allocated storage.

D.Create a new DB instance with larger storage and migrate the data.

AnswerC

RDS allows online storage modification.

Why this answer

Amazon RDS allows you to modify allocated storage online using the 'Modify DB Instance' action without significant downtime. Option A is wrong because deleting old archived redo logs can free up space temporarily but does not add storage and may impact point-in-time recovery. Option B is wrong because taking a snapshot and restoring to a larger instance requires downtime during the restore process.

Option D is wrong because creating a new instance and migrating data involves downtime and is more complex than simply modifying storage.

Practice this question →

219

MCQhard

A company uses Amazon RDS for MySQL with read replicas. The application writes to the primary and reads from the replicas. Occasionally, the application reads stale data from the replicas. Which action would ensure read-after-write consistency without impacting write performance?

A.Set the replica_read_consistency parameter to 'session'.

B.Use the reader endpoint and increase the replica lag threshold.

C.Configure the application to read from the primary instance for critical queries.

D.Enable the 'rds_set_replication_status' parameter on the read replicas.

AnswerC

Reading from primary guarantees consistency.

Why this answer

Using the primary instance for reads that require consistency ensures the application reads the latest data. Session-level replication checks are not supported in standard MySQL.

Practice this question →

220

MCQeasy

A company has an Amazon DynamoDB table with on-demand capacity mode. The table is used by a serverless application. The company wants to receive an alert when the read request rate exceeds a certain threshold. Which CloudWatch metric and alarm should be used?

A.Alarm on the 'ConsumedReadCapacityUnits' metric.

B.Alarm on the 'ReadThrottleEvents' metric.

C.Alarm on the 'ProvisionedReadCapacityUnits' metric.

D.Alarm on the 'SuccessfulRequestLatency' metric.

AnswerA

This metric shows actual read usage.

Why this answer

The 'ConsumedReadCapacityUnits' metric reflects the actual number of read capacity units consumed by the table. By setting an alarm on this metric, you can trigger an alert when the read request rate exceeds a defined threshold, which is appropriate for an on-demand table where capacity scales automatically but you still want to monitor usage.

Exam trap

The trap here is that candidates often confuse throttling events (ReadThrottleEvents) with actual consumption, but the question asks for an alert when the read request rate exceeds a threshold, which requires monitoring consumption, not throttling.

How to eliminate wrong answers

Option B is wrong because 'ReadThrottleEvents' measures throttled read requests, not the request rate itself; it would only alert after throttling occurs, not before exceeding a threshold. Option C is wrong because 'ProvisionedReadCapacityUnits' is only applicable to provisioned capacity mode, not on-demand mode, and would always be zero or irrelevant. Option D is wrong because 'SuccessfulRequestLatency' measures response time, not request rate, and is not suitable for alerting on throughput thresholds.

Practice this question →

221

MCQhard

A company runs a critical Amazon RDS for PostgreSQL database. They notice that the 'DiskQueueDepth' metric is consistently high and the 'FreeStorageSpace' is below 10%. The database is used for OLTP workloads. What is the MOST immediate action to take?

A.Upgrade to a larger DB instance class.

B.Modify the DB instance to increase allocated storage.

C.Switch the instance to Provisioned IOPS storage.

D.Create a read replica to offload read traffic.

AnswerB

Increasing storage immediately addresses the low free space and can reduce disk queue depth by allowing more I/O operations.

Why this answer

Increasing allocated storage quickly addresses the low free storage space (below 10%) and high DiskQueueDepth, which indicates I/O bottlenecks due to insufficient storage. Option A is wrong because upgrading to a larger instance class improves memory and CPU but does not directly resolve storage shortage or I/O throttling caused by low disk space. Option C is wrong because switching to Provisioned IOPS improves latency and throughput but does not increase storage capacity.

Option D is wrong because creating a read replica offloads read traffic but does not increase storage on the primary instance.

Practice this question →

222

MCQmedium

An Amazon RDS for Oracle DB instance is experiencing high swap usage. The database administrator wants to reduce swap usage. Which action should be taken?

A.Change the storage type from gp2 to io1

B.Increase the DB instance class to a larger size with more memory

C.Delete archived redo logs to free up space

D.Enable Multi-AZ to distribute the load

AnswerB

More memory reduces the need for swap.

Why this answer

High swap usage on an Amazon RDS for Oracle DB instance indicates that the operating system is using disk-based swap space as a substitute for physical RAM, which severely degrades database performance. Increasing the DB instance class to a larger size with more memory directly addresses the root cause by providing additional RAM, reducing or eliminating the need for swapping. This is the correct action because swap usage is a memory pressure issue, not a storage or availability problem.

Exam trap

The trap here is that candidates confuse high swap usage with a storage performance or availability issue, leading them to choose storage type changes or Multi-AZ, when the real solution is to address insufficient memory by scaling the instance class.

How to eliminate wrong answers

Option A is wrong because changing the storage type from gp2 to io1 improves I/O performance and latency but does not increase the amount of available RAM, so it cannot reduce swap usage. Option C is wrong because deleting archived redo logs frees up storage space in the recovery area, not memory; swap usage is unrelated to log retention or disk space. Option D is wrong because enabling Multi-AZ provides high availability and automatic failover by replicating data to a standby instance, but it does not distribute memory load or reduce swap usage on the primary instance.

Practice this question →

223

MCQhard

The exhibit shows an IAM policy attached to a user. The user needs to create a manual snapshot of an RDS DB instance named 'production-db'. Which action will the user be able to perform?

A.Create a manual snapshot of 'production-db' with the name 'production-db-snapshot'.

B.Create a manual snapshot of 'production-db' with the name 'mydb-production-snapshot'.

C.Describe the 'production-db' DB instance.

D.Delete the 'production-db' DB instance.

AnswerB

The snapshot name starts with 'mydb-', matching the allowed resource pattern.

Why this answer

The IAM policy allows CreateDBSnapshot only on DB instances with names starting with 'mydb-', and the snapshot name must also match the pattern 'mydb-*'. The snapshot name 'mydb-production-snapshot' satisfies that pattern. Option A is wrong because 'production-db-snapshot' does not start with 'mydb-'.

Option C is wrong because the question asks about creating a snapshot, not describing instances. Option D is wrong because the policy does not grant DeleteDBInstance permission.

Practice this question →

224

MCQhard

A company runs a production Amazon RDS for MySQL DB instance with Multi-AZ. The database is used by a web application. The application team reports that the database is experiencing intermittent connection timeouts and increased latency. The CloudWatch metrics show that the database connections spike to the maximum allowed (max_connections) during peak hours, and the CPU utilization is high. The team needs to resolve the connection issues without modifying the application code. The application uses connection pooling at the application layer. Which action should be taken?

A.Create a read replica and direct read queries to it.

B.Deploy an Amazon RDS Proxy in front of the DB instance.

C.Increase the 'max_connections' parameter to allow more connections.

D.Change the DB instance class to a smaller size to reduce the maximum connections.

AnswerB

Amazon RDS Proxy efficiently manages connection pooling, reducing the number of connections to the database and lowering CPU overhead, thus resolving connection timeouts without code changes.

Why this answer

RDS Proxy manages connection pooling efficiently, reducing the number of connections to the database and lowering CPU overhead. Option A is wrong because read replicas do not reduce write connection load. Option C is wrong because increasing max_connections may lead to resource exhaustion.

Option D is wrong because switching to a smaller instance would worsen the problem.

Practice this question →

225

Multi-Selecthard

A company is migrating an on-premises Oracle database to Amazon RDS for Oracle. The database is 2 TB in size and has a tight migration window of 8 hours. Which THREE steps should be taken to minimize downtime during the migration?

Select 3 answers

A.Take a manual snapshot of the source database before migration.

B.Use AWS Schema Conversion Tool (AWS SCT) to convert the schema.

C.Use AWS Database Migration Service (AWS DMS) with ongoing replication.

D.Disable automated backups on the target RDS instance during migration.

E.Deploy a large RDS instance class to speed up the initial load.

AnswersC, D, E

Correct. AWS Database Migration Service (AWS DMS) with ongoing replication allows continuous replication from source to target, minimizing downtime during the final cutover.

Why this answer

AWS DMS with ongoing replication (option C) allows continuous data replication from the source to the target, minimizing downtime during the final cutover. Disabling automated backups on the target RDS instance (option D) reduces I/O overhead and speeds up the migration process. Deploying a large RDS instance class (option E) provides higher network and disk throughput, accelerating the initial data load.

Option A is incorrect because a manual snapshot is not used for migration, and option B is incorrect because AWS SCT is for schema conversion, not data replication.

Practice this question →

← PreviousPage 3 of 4 · 300 questions totalNext →

Ready to test yourself?

Try a timed practice session using only Db Mgmt Ops questions.

Start 20-question session