Courseiva

AWS Certified Database Specialty DBS-C01 (DBS-C01) — Questions 14261500

1663 questions total · 23pages · All types, answers revealed

Page 19

Page 20 of 23

Page 21
1426
MCQhard

A company uses Amazon DynamoDB for a shopping cart application. During a flash sale, write requests are throttled due to hot partitions. The access pattern is evenly distributed across items, but the partition key is the customer ID. Which design change would best mitigate throttling?

A.Enable DynamoDB adaptive capacity.
B.Change the partition key to a single value for all items.
C.Increase the provisioned write capacity to a higher fixed value.
D.Add a DAX cluster in front of DynamoDB.
AnswerA

Adaptive capacity rebalances throughput across partitions.

Why this answer

DynamoDB adaptive capacity automatically adjusts throughput capacity based on traffic patterns, which helps mitigate hot partitions by redistributing unused capacity from less-accessed partitions to heavily accessed ones. This is ideal for the flash sale scenario where write requests are throttled due to uneven access across customer ID partitions, even though the overall access pattern is evenly distributed.

Exam trap

The trap here is that candidates may think increasing provisioned capacity (Option C) is the straightforward fix for throttling, but they overlook that hot partitions require a design-level solution like adaptive capacity or partition key redesign to distribute writes evenly.

How to eliminate wrong answers

Option B is wrong because changing the partition key to a single value for all items would create an extreme hot partition, causing all writes to target one partition and severely throttling the entire table. Option C is wrong because increasing provisioned write capacity to a higher fixed value does not address the root cause of hot partitions; it only increases overall throughput but still allows throttling on individual partitions if the access pattern is skewed. Option D is wrong because adding a DAX cluster in front of DynamoDB is a caching layer that primarily improves read performance and reduces read latency, but it does not mitigate write throttling or hot partition issues on the write path.

1427
MCQhard

A company runs a critical Oracle database on Amazon RDS. The database has a large table that is frequently accessed by multiple applications. The team wants to implement caching to reduce the load on the database. The cached data must be strongly consistent with the database. Which caching strategy should they use?

A.Eventual consistency with DynamoDB Accelerator (DAX)
B.Read-only cache with Amazon ElastiCache
C.Write-through cache using Amazon ElastiCache
D.Lazy loading with cache-aside pattern
AnswerC

Write-through ensures data is written to cache and DB together, maintaining strong consistency.

Why this answer

The write-through cache strategy ensures that every write to the database also updates the cache synchronously, so the cached data is always strongly consistent with the database. This is critical for the Oracle RDS workload where multiple applications require immediate consistency. Amazon ElastiCache (Redis or Memcached) supports write-through by updating the cache on every write operation, preventing stale reads.

Exam trap

The trap here is that candidates often confuse 'read-only cache' or 'lazy loading' with strong consistency, not realizing that only write-through synchronously updates the cache on every write, making it the only option that guarantees the cached data is always identical to the database.

How to eliminate wrong answers

Option A is wrong because DynamoDB Accelerator (DAX) is an in-memory cache for DynamoDB, not for Amazon RDS Oracle, and eventual consistency does not meet the strong consistency requirement. Option B is wrong because a read-only cache only caches data that is read, not written, so it cannot ensure strong consistency for data that is updated; it would serve stale data until the cache is invalidated or refreshed. Option D is wrong because lazy loading (cache-aside) loads data into the cache only on a cache miss, which can lead to stale data if the database is updated before the cache is refreshed; it does not guarantee strong consistency.

1428
Multi-Selecthard

Refer to the exhibit. A database engineer is assigned this IAM policy. Which of the following actions can the engineer perform? (Choose two.)

Select 2 answers
A.Describe all automated snapshots
B.Delete a manual snapshot named dev-snapshot
C.Delete a manual snapshot named prod-database-snapshot
D.Delete a manual snapshot named prod-backup
E.Create a manual snapshot named test-snapshot
AnswersA, E

The Allow statement permits DescribeDBSnapshots for all resources.

Why this answer

The IAM policy includes an Allow statement for 'rds:DescribeDBSnapshots' on all resources, so the engineer can describe any snapshot, including automated snapshots (Option A). It also includes an Allow statement for 'rds:CreateDBSnapshot' on all resources, allowing creation of a manual snapshot with any name, such as 'test-snapshot' (Option E). There is no Allow statement for 'rds:DeleteDBSnapshot', and a Deny statement explicitly blocks deletion of snapshots with names starting with 'prod-'.

Therefore, deleting any manual snapshot (Options B, C, D) is not permitted. The correct answers are A and E.

1429
MCQmedium

A company is designing a database for a social media application that stores user posts. Each post can have multiple tags. The workload requires low-latency queries to find all posts with a specific tag. Which database design is most suitable?

A.Amazon ElastiCache for Memcached storing posts and tags as key-value pairs.
B.Amazon DynamoDB with a Global Secondary Index on the tag attribute.
C.Amazon RDS for MySQL with a normalized schema and JOIN queries.
D.Amazon Neptune with a graph model for tags and posts.
AnswerB

GSI provides fast query by tag.

Why this answer

Amazon DynamoDB with a Global Secondary Index (GSI) on the tag attribute is the most suitable design because it allows low-latency queries to find all posts with a specific tag without scanning the entire table. The GSI enables efficient querying by tag as a partition key, supporting the required access pattern with consistent single-digit millisecond performance at any scale.

Exam trap

The trap here is that candidates may choose Amazon RDS for MySQL (Option C) due to familiarity with normalized relational designs, overlooking that DynamoDB's GSI provides superior performance and scalability for high-velocity, low-latency tag-based queries without the overhead of JOINs.

How to eliminate wrong answers

Option A is wrong because Amazon ElastiCache for Memcached is an in-memory cache, not a persistent database; it lacks native indexing for tag-based queries and would require application-level logic to maintain and query tag-to-post mappings, leading to data loss on cache eviction or failure. Option C is wrong because Amazon RDS for MySQL with a normalized schema and JOIN queries introduces relational overhead and potential performance bottlenecks at scale, as JOINs on large tables with many-to-many relationships (posts and tags) cannot match the low-latency, single-query access provided by DynamoDB's GSI. Option D is wrong because Amazon Neptune, while capable of modeling tags and posts as a graph, is overkill for this simple key-value access pattern and incurs higher latency and cost compared to DynamoDB's direct index lookup.

1430
MCQmedium

A company runs an e-commerce platform on Amazon RDS for PostgreSQL. During a flash sale, the database experiences high write load and read replicas lag significantly. The application uses read replicas for reporting queries. Which design change would most effectively reduce replica lag without compromising write performance?

A.Increase the instance size of the primary database.
B.Add more read replicas to distribute the reporting load.
C.Migrate to Amazon Aurora with read replicas.
D.Convert the RDS instance to a Multi-AZ deployment.
AnswerC

Aurora has faster replication (typically <100ms) and is designed to handle high write loads with minimal replica lag.

Why this answer

Amazon Aurora's distributed storage architecture decouples compute from storage, allowing replicas to apply redo logs with minimal overhead compared to RDS for PostgreSQL's physical replication. Aurora's replicas share the same underlying storage volume, so replica lag is significantly reduced even under heavy write loads, while write performance on the primary remains unaffected due to the asynchronous, log-based replication mechanism.

Exam trap

The trap here is that candidates assume adding more replicas or scaling the primary will solve replication lag, but they fail to recognize that the fundamental replication mechanism in RDS for PostgreSQL (streaming WAL) is the bottleneck, whereas Aurora's shared-storage architecture inherently minimizes lag.

How to eliminate wrong answers

Option A is wrong because increasing the primary instance size may improve write throughput but does not address the root cause of replica lag, which is the replication bottleneck in RDS for PostgreSQL's streaming replication; the primary's larger size does not speed up log shipping or apply on replicas. Option B is wrong because adding more read replicas does not reduce lag on existing replicas; it may even increase replication overhead on the primary, potentially worsening lag for all replicas under high write load. Option D is wrong because Multi-AZ deployment provides high availability with synchronous replication to a standby instance, but it does not create read replicas or reduce lag for reporting queries; the standby is not used for reads and does not alleviate replica lag.

1431
MCQeasy

A company is using Amazon RDS for Oracle and wants to integrate with AWS CloudTrail to log database API calls. Which action is necessary?

A.Enable CloudTrail for the RDS instance.
B.Create a VPC endpoint for CloudTrail.
C.Configure Oracle Fine-Grained Auditing (FGA).
D.Install the pgAudit extension.
AnswerA

CloudTrail already logs RDS API calls; no special setup needed.

Why this answer

AWS CloudTrail is the service that logs API calls made to AWS services, including Amazon RDS. To capture database API calls (e.g., CreateDBInstance, ModifyDBInstance) for an RDS for Oracle instance, you must enable CloudTrail for the RDS instance by creating a trail that covers the RDS service. This logs management events at the AWS control plane level, not the database engine level.

Exam trap

The trap here is confusing AWS-level API logging (CloudTrail) with database engine-level auditing (FGA, pgAudit), leading candidates to select database-specific auditing tools instead of the correct AWS service for logging control plane API calls.

How to eliminate wrong answers

Option B is wrong because a VPC endpoint for CloudTrail is used to privately connect your VPC to CloudTrail without using the public internet, but it is not required to log RDS API calls; CloudTrail works over the public AWS API endpoints by default. Option C is wrong because Oracle Fine-Grained Auditing (FGA) is a database-level auditing feature that logs SQL operations within the Oracle database engine, not AWS API calls to the RDS service. Option D is wrong because the pgAudit extension is used for PostgreSQL databases to log database-level activity, not for Oracle RDS instances or AWS API logging.

1432
MCQmedium

A company is migrating an on-premises PostgreSQL database to Amazon RDS for PostgreSQL. The database is 2 TB and has a sustained write rate of 150 MB/s. The migration must have minimal downtime. Which AWS service or tool should be used for the initial data load and ongoing replication?

A.Use AWS Database Migration Service (AWS DMS) with full load and ongoing replication
B.Use pg_dump and pg_restore
C.Use AWS Schema Conversion Tool (AWS SCT) to convert schema and AWS DMS for data migration
D.Use AWS S3 Transfer Acceleration to upload data to Amazon S3, then restore to RDS
AnswerA

AWS DMS supports full load and ongoing replication with minimal downtime.

Why this answer

AWS DMS is the correct choice because it supports both a full load of the existing 2 TB database and ongoing change data capture (CDC) to replicate transactions with minimal downtime. DMS can handle sustained write rates of 150 MB/s by using appropriate instance sizing and tuning, and it provides built-in mechanisms for resumable loads and validation, which are critical for a large-scale migration with minimal downtime.

Exam trap

The trap here is that candidates may assume pg_dump/pg_restore can be used with minimal downtime by running them on a replica, but they still require a consistent snapshot and cannot capture ongoing changes, making them unsuitable for near-zero downtime migrations.

How to eliminate wrong answers

Option B is wrong because pg_dump and pg_restore are offline tools that require the source database to be stopped or made read-only during the dump, which does not meet the minimal downtime requirement; they also cannot perform ongoing replication. Option C is wrong because AWS SCT is used for schema conversion when migrating between different database engines (e.g., Oracle to PostgreSQL), but the question specifies a homogeneous migration from on-premises PostgreSQL to RDS for PostgreSQL, so no schema conversion is needed. Option D is wrong because S3 Transfer Acceleration only speeds up uploads to S3, but restoring from S3 to RDS would still require a full offline restore and cannot provide ongoing replication; it also adds unnecessary complexity and latency for a direct database migration.

1433
MCQhard

An IAM policy is attached to a user who is deploying a new RDS instance. What is the effect of this policy on the user's ability to modify an existing production database instance with the identifier 'prod-mydb'?

A.The user cannot modify the production database instance because the Deny statement explicitly denies ModifyDBInstance on production databases.
B.The user can modify the production database instance because the Allow statement grants full access to RDS actions.
C.The user can modify the production database instance if they use the AWS CLI instead of the console.
D.The user cannot modify any database instance because the Deny statement denies ModifyDBInstance on all resources.
AnswerA

Deny overrides Allow, and the resource pattern matches the production database.

Why this answer

The IAM policy includes an explicit Deny statement that denies the `rds:ModifyDBInstance` action when the resource condition matches `arn:aws:rds:*:*:db:prod-mydb`. In IAM, an explicit Deny overrides any Allow, so even though the Allow statement grants full RDS access, the Deny takes precedence and blocks modification of the production database instance with identifier 'prod-mydb'.

Exam trap

The trap here is that candidates often assume an Allow statement with full access will override a Deny, but AWS IAM explicitly prioritizes Deny over Allow, and the Deny's resource condition restricts the effect to only the named production database, not all databases.

How to eliminate wrong answers

Option B is wrong because it ignores the explicit Deny statement; in IAM, an explicit Deny always overrides an Allow, so the Allow statement does not grant the ability to modify the specified production database. Option C is wrong because IAM policies are service-agnostic and apply equally to the AWS Management Console, CLI, and SDKs; the Deny statement blocks the action regardless of the interface used. Option D is wrong because the Deny statement is scoped to the specific resource `arn:aws:rds:*:*:db:prod-mydb`, not to all database instances; the user can still modify other RDS instances that do not match that resource ARN.

1434
MCQmedium

Refer to the exhibit. An IAM policy is attached to a user. The user attempts to delete a DB instance named 'prod-db'. What will happen?

A.The delete will succeed because the Allow statement grants modify permission.
B.The delete will succeed because the resource is 'prod-db' which does not match the deny pattern.
C.The delete will fail because of the explicit Deny statement.
D.The delete will succeed only if the user has MFA enabled.
AnswerC

Explicit deny overrides any allow.

Why this answer

The Deny statement explicitly denies delete on any instance matching 'prod-*'. Since an explicit deny overrides any allow, the delete will fail. Therefore, the correct answer is C.

Option A is incorrect because the deny overrides the allow. Option B is incorrect because the resource 'prod-db' does match the deny pattern 'prod-*'. Option D is incorrect because the policy does not require MFA in this context, and even with MFA, the explicit deny would still apply.

1435
MCQmedium

A company uses Amazon DynamoDB for a time-series IoT application. Each device sends a data point every second. The application queries data by device ID and timestamp range. Which table design is most efficient?

A.Use a composite key of device ID and timestamp as the partition key.
B.Use device ID as the partition key and a random suffix as the sort key.
C.Use timestamp as the partition key and device ID as the sort key.
D.Use device ID as the partition key and timestamp as the sort key.
AnswerD

Allows efficient range queries on timestamp per device.

Why this answer

It models the access pattern directly: using device ID as the partition key ensures all data for a device is co-located, and timestamp as the sort key enables efficient range queries (e.g., Query with KeyConditionExpression on timestamp between start and end). This design avoids hot partitions and allows DynamoDB to retrieve the exact time-series slice without scanning.

Exam trap

AWS often tests the misconception that a composite partition key (device ID + timestamp) is needed for uniqueness, but the trap here is that candidates forget the sort key's role in enabling range queries and instead try to force uniqueness into the partition key, which breaks the access pattern.

How to eliminate wrong answers

Option A is wrong because using a composite key of device ID and timestamp as the partition key would create a unique partition for each data point, making it impossible to query all data for a device across a time range without a full scan. Option B is wrong because using a random suffix as the sort key destroys the natural ordering of timestamps, preventing efficient range queries and forcing a scan to filter by time. Option C is wrong because using timestamp as the partition key leads to a single hot partition for each second (or time granularity), causing throttling and poor distribution, and querying by device ID would require a scan across all partitions.

1436
MCQmedium

Refer to the exhibit. A company is migrating an on-premises database to Amazon RDS for MySQL. During a test migration, the DMS task fails with the error shown. The source database is 500 GB and the target RDS instance has 500 GB allocated storage. What should be done to resolve this error and complete the migration?

A.Increase the DB instance class to a larger size
B.Increase the DMS task's allocated storage
C.Reduce the size of the source database
D.Enable storage autoscaling on the RDS instance
AnswerD

Autoscaling will automatically increase storage as needed.

Why this answer

The error indicates that the target RDS instance ran out of storage during the migration. Enabling storage autoscaling allows RDS to automatically increase storage when free space is low, preventing the migration from failing due to insufficient disk space. This is the correct resolution because the source database is 500 GB and the target has exactly 500 GB allocated, leaving no room for temporary data or logs during the migration.

Exam trap

The trap here is that candidates confuse storage allocation with instance class or DMS storage, thinking performance upgrades or DMS settings will fix a disk space issue, when the real problem is simply that the target RDS instance needs more storage capacity.

How to eliminate wrong answers

Option A is wrong because increasing the DB instance class (CPU/memory) does not increase storage capacity; it addresses performance, not disk space. Option B is wrong because DMS tasks do not have allocated storage; DMS uses replication instances with their own storage, but the error is about the target RDS instance's storage, not DMS storage. Option C is wrong because reducing the source database size is unnecessary and impractical; the migration should handle the full dataset, and the issue is insufficient target storage, not source size.

1437
Multi-Selectmedium

A company runs a self-managed Redis cluster on Amazon EC2 for caching. The cluster has one primary and two replicas, each on c5.large instances. The application experiences high latency during peak hours. CloudWatch metrics show that the primary node's CPU utilization is consistently above 80% and the network bandwidth is near the instance limit. The replicas show moderate CPU usage. The team wants to reduce latency without increasing cost significantly. Which combination of actions should the team take? (Choose two.)

Select 2 answers
A.Enable Redis Cluster Mode and distribute data across multiple shards.
B.Add more EC2 instances as additional replicas to offload reads.
C.Configure the application to read from replica nodes.
D.Upgrade the primary to a c5.2xlarge instance type.
E.Migrate to Amazon ElastiCache for Redis with Cluster Mode enabled.
AnswersA, E

Correct. Enabling Redis Cluster Mode shards data across multiple nodes, reducing CPU and network load on the primary.

Why this answer

Enables Redis Cluster Mode to shard data across multiple nodes, reducing CPU and network load on the primary. Option E migrates to Amazon ElastiCache for Redis with Cluster Mode, which manages scaling and reduces operational overhead. Together, they address the bottleneck without significant cost increase.

Option C (read from replicas) does not reduce primary write/CPU load. Option D (upgrade instance) increases cost and may still hit network limits. Option B (add replicas) increases cost and does not reduce primary load.

Exam trap

Candidates may incorrectly select reading from replicas (option C) assuming it alleviates the primary's CPU, but the primary's high CPU is due to both CPU and network saturation from read/write operations; read replicas only offload read queries, not the underlying processing on the primary.

1438
Multi-Selecteasy

Which THREE actions should be taken to troubleshoot an Amazon RDS for PostgreSQL instance that is unresponsive? (Choose 3.)

Select 3 answers
A.Reboot the DB instance immediately
B.Modify the DB instance to a larger instance class
C.Verify that the security group allows inbound traffic on the database port
D.Check the database error logs in CloudWatch Logs
E.Review CloudWatch metrics for CPU, memory, and disk I/O
AnswersC, D, E

Network connectivity issues can make instance appear unresponsive.

Why this answer

To troubleshoot an unresponsive Amazon RDS for PostgreSQL instance, the standard initial steps are: C) Verify that the security group allows inbound traffic on the database port – connectivity issues are a common cause of unresponsiveness. D) Check the database error logs in CloudWatch Logs – logs may reveal database-specific errors like out-of-memory or corruption. E) Review CloudWatch metrics for CPU, memory, and disk I/O – high utilization can indicate resource exhaustion.

Option A (reboot immediately) is wrong because it may destroy diagnostic evidence. Option B (modify instance class) is premature without first assessing the problem.

1439
MCQeasy

A company is using Amazon DynamoDB and wants to ensure that only authorized users can access a specific table. Which AWS service should be used to manage access control?

A.AWS CloudHSM.
B.AWS Key Management Service (KMS).
C.Amazon VPC security groups.
D.AWS Identity and Access Management (IAM).
AnswerD

Correct. IAM allows you to create policies that define who can access which DynamoDB tables and what actions they can perform.

Why this answer

AWS Identity and Access Management (IAM) is the service used to manage access control for DynamoDB tables by creating policies that grant or deny permissions. Option A (CloudHSM) provides hardware security modules for encryption keys, not access control. Option B (KMS) manages encryption keys but does not handle access control.

Option C (VPC security groups) control network traffic at the instance level, not database-level access control.

1440
MCQeasy

A financial services company needs a relational database with high availability and automatic failover across three Availability Zones in us-east-1. The workload consists of OLTP transactions with occasional analytic queries. Which database solution meets these requirements?

A.Amazon RDS for MySQL with Multi-AZ (2 AZs)
B.Amazon DynamoDB global tables
C.Amazon Aurora MySQL with Multi-AZ deployment across 3 AZs
D.Amazon Redshift with cross-region snapshots
AnswerC

Aurora provides automatic failover across 3 AZs and supports OLTP and analytics.

Why this answer

Amazon Aurora MySQL with Multi-AZ deployment across 3 AZs meets the requirements because Aurora automatically replicates your data six ways across three Availability Zones, with a primary DB instance in one AZ and two read replicas in the other two AZs. In the event of a failure, Aurora automatically fails over to a read replica in under 30 seconds without data loss, providing high availability and automatic failover across three AZs. Aurora also supports both OLTP transactions and can handle occasional analytic queries via Aurora Replicas or Aurora Global Database for read scaling.

Exam trap

The trap here is that candidates often confuse RDS Multi-AZ (which only supports 2 AZs) with Aurora's native multi-AZ replication across 3 AZs, or they mistakenly think DynamoDB global tables (a NoSQL service) can replace a relational database for OLTP workloads requiring ACID transactions.

How to eliminate wrong answers

Option A is wrong because Amazon RDS for MySQL with Multi-AZ (2 AZs) only supports a standby replica in a single secondary AZ, not across three AZs, and failover is limited to two AZs, failing the requirement for three Availability Zones. Option B is wrong because Amazon DynamoDB global tables is a NoSQL key-value and document database, not a relational database, and while it provides multi-region replication, it does not support relational queries or ACID transactions in the same way as a relational database, and it does not offer automatic failover across three AZs in a single region. Option D is wrong because Amazon Redshift is a data warehouse optimized for analytic queries, not OLTP transactions, and cross-region snapshots provide disaster recovery but not automatic failover across three AZs for high availability.

1441
Multi-Selecthard

A company is migrating an on-premises Oracle database to Amazon RDS for Oracle. The database contains sensitive personally identifiable information (PII). The security team requires that the data be encrypted in transit between the application and the RDS instance, and also encrypted at rest using a key managed by the company. Which THREE actions should the company take? (Choose THREE.)

Select 3 answers
A.Configure the RDS instance to require SSL/TLS connections.
B.Modify the DB parameter group to set the 'rds.force_ssl' parameter to 1.
C.Enable encryption at rest for the RDS instance at launch time.
D.Enable Transparent Data Encryption (TDE) using CloudHSM.
E.Create a customer managed key (CMK) in AWS KMS.
AnswersA, C, E

SSL/TLS encrypts data in transit.

Why this answer

Options A, C, and E are correct. Option A: Configuring the RDS instance to require SSL/TLS encrypts data in transit, meeting the encryption-in-transit requirement. Option C: Enabling encryption at rest for the RDS instance at launch time allows use of AWS KMS for encryption, satisfying the at-rest encryption requirement.

Option E: Creating a customer managed key (CMK) in AWS KMS enables the company to manage the encryption key, meeting the requirement for a key managed by the company. Option B is incorrect because setting rds.force_ssl to 1 in the DB parameter group is a step within configuring SSL/TLS, but the primary action is option A; moreover, the correct method to enforce SSL for Oracle RDS involves using the option group with the SSL option, not just the parameter group. Option D is incorrect because Transparent Data Encryption (TDE) using CloudHSM is not required; the at-rest encryption requirement is already fulfilled by enabling RDS encryption with a CMK (options C and E), and TDE with CloudHSM adds unnecessary complexity and potential licensing issues.

1442
MCQmedium

A company uses Amazon RDS for PostgreSQL with logical replication to a downstream system. The replication slot grows unbounded and causes storage full issues. Which action resolves this without data loss?

A.Disable logical replication and use DMS instead
B.Increase the allocated storage size of the RDS instance
C.Monitor the replication slot and advance it using pg_replication_slot_advance
D.Delete the replication slot and recreate it
AnswerC

Advancing the slot allows WAL cleanup.

Why this answer

Monitoring and advancing the replication slot prevents WAL accumulation. Option A is wrong because disabling replication causes data loss. Option B is wrong because increasing storage is a temporary fix, not a resolution.

Option D is wrong because deleting old WAL logs may break replication.

1443
MCQhard

An e-commerce company runs a multi-AZ deployment of Amazon RDS for MySQL. During a recent failover test, the application experienced a 30-second write outage. The application uses a connection pooling library. The DB instance has a 60-second TTL for DNS records. What is the MOST likely cause of the outage?

A.The connection pool had open connections to the old primary, and DNS TTL caused a delay in reconnecting to the new primary.
B.The application experienced a cold start after the failover.
C.The DNS record for the RDS endpoint was not updated after the failover.
D.The Multi-AZ failover took longer than 30 seconds to complete.
AnswerA

Stale connections and DNS caching can cause a brief outage until connections are refreshed.

Why this answer

During a Multi-AZ failover, the RDS DNS record is automatically updated to point to the new primary. However, the application's connection pool may still have open connections to the old primary IP. Because the DNS TTL is 60 seconds, the client may continue to resolve to the old (cached) IP for up to 60 seconds, causing a write outage until connections are re-established to the new primary.

Option B is incorrect because a cold start typically refers to an application starting from scratch, which is not the case here. Option C is incorrect because the DNS record is indeed updated after failover; the issue is client-side caching. Option D is incorrect because Multi-AZ failover usually completes within 1-2 minutes, but the outage duration is determined by DNS TTL and connection pooling behavior, not the failover time itself.

1444
MCQmedium

A company is migrating a 3 TB PostgreSQL database from on-premises to Amazon Aurora PostgreSQL. They need to minimize downtime and ensure that the migration is completed within a maintenance window. Which approach should they use?

A.Use AWS DMS with full load and ongoing replication (CDC)
B.Use AWS SCT to convert the schema and then migrate data
C.Use pg_dump to export the database and pg_restore to import into Aurora
D.Take a file system snapshot and restore to Aurora
AnswerA

DMS with CDC supports minimal downtime by replicating changes.

Why this answer

AWS DMS with full load and ongoing replication (CDC) is the correct approach because it allows you to perform an initial full load of the 3 TB database while continuously capturing changes from the source PostgreSQL using logical replication (via the pglogical extension or native slot-based replication). This minimizes downtime by keeping the target Aurora PostgreSQL nearly synchronized, and you can switch over during a maintenance window with only a brief outage to apply any final lag.

Exam trap

The trap here is that candidates often choose pg_dump/pg_restore (Option C) because it is a familiar tool, but they overlook the requirement to minimize downtime, which CDC-based replication uniquely addresses.

How to eliminate wrong answers

Option B is wrong because AWS SCT is used for schema conversion (e.g., from Oracle or SQL Server to Aurora PostgreSQL), not for migrating data from PostgreSQL to PostgreSQL, and it does not provide ongoing replication to minimize downtime. Option C is wrong because pg_dump/pg_restore is a logical backup and restore method that requires the source database to be offline or read-only during the dump, causing significant downtime for a 3 TB database, and it does not support continuous replication. Option D is wrong because file system snapshots are not compatible with Aurora PostgreSQL; Aurora uses a distributed storage layer and cannot ingest raw file system snapshots from on-premises PostgreSQL.

1445
MCQhard

A company uses Amazon DynamoDB with On-Demand capacity for a gaming application. During a new game launch, write traffic spikes 10x normal for 30 minutes. Some write requests receive ProvisionedThroughputExceeded exceptions. What is the MOST likely cause and solution?

A.The partition key is not distributing writes evenly, causing a hot partition. Redesign the partition key for uniform access.
B.On-Demand capacity cannot handle sudden spikes. Switch to Provisioned capacity with auto scaling.
C.DynamoDB Streams is enabled, causing additional write throttling. Disable streams.
D.The table has a global secondary index with a different partition key that is unevenly accessed. Remove the GSI.
AnswerA

Hot partitions cause throttling even with On-Demand capacity.

Why this answer

On-Demand capacity can handle spikes but has per-partition throughput limits. If a hot partition exists, writes to that partition may exceed its limit. Option A is correct because uneven partition keys cause throttling.

Option B is incorrect because On-Demand does not have table-level limits. Option C is not the primary cause. Option D is incorrect as GSI writes also consume write capacity.

1446
MCQmedium

A company is running an Amazon RDS for MySQL Multi-AZ DB instance. The application experiences a brief write disruption during automatic failover. The database workload has low write latency requirements. Which configuration change would minimize application impact during failover?

A.Create a read replica in a different Availability Zone and promote it during failover.
B.Change the DB instance to use a Multi-AZ DB cluster configuration.
C.Use a single-AZ deployment and rely on automated backups for recovery.
D.Increase the DB instance class to reduce failover time.
AnswerB

Multi-AZ DB cluster provides faster failover than standard Multi-AZ.

Why this answer

Changing to a Multi-AZ DB cluster (option B) reduces application impact during failover because it uses a writer and two reader instances in separate Availability Zones. Failover to a reader is typically faster than the standard Multi-AZ standby failover, minimizing write disruption. Option A (promoting a read replica) is not automatic and introduces significant delay.

Option C (single-AZ with backups) does not address failover impact. Option D (increasing instance class) does not significantly reduce failover time.

1447
Multi-Selectmedium

A company is using Amazon DynamoDB with autoscaling enabled. The table has a partition key of 'order_id' and a sort key of 'order_date'. The application performs both point queries and range queries. Recently, the 'ConsumedReadCapacityUnits' metric shows that the table is consistently using 100% of the provisioned capacity. Which THREE factors should the database engineer investigate to determine the cause?

Select 3 answers
A.Whether autoscaling is configured correctly to add capacity.
B.Whether the application is using Scan operations instead of Query operations.
C.Whether the partition key is evenly distributed across partitions.
D.Whether a specific 'order_id' is being accessed frequently, creating a hot key.
E.Whether a global secondary index is being used for queries.
AnswersB, C, D

Scans consume more read capacity than queries.

Why this answer

Scan operations read the entire table or index before applying filters, consuming far more read capacity than Query operations, which target specific partition and sort key values. If the application is using Scans instead of Queries, it would consistently consume 100% of provisioned capacity even for small result sets, leading to throttling and high utilization.

Exam trap

AWS often tests the misconception that autoscaling misconfiguration is the primary cause of high capacity utilization, when in reality the root cause is often inefficient access patterns (Scans) or uneven data distribution (hot keys) that autoscaling cannot fix.

1448
MCQhard

A company is running a production Amazon DynamoDB table that supports a gaming application with millions of concurrent users. The table uses on-demand capacity mode. Recently, the application started experiencing throttling (ProvisionedThroughputExceededException) during peak hours. The company wants to resolve this with minimal operational overhead. What should the company do?

A.Enable DynamoDB Accelerator (DAX) to cache reads
B.Partition the table across multiple tables and use application-level sharding
C.Request a service quota increase for the on-demand table's maximum throughput
D.Switch to provisioned capacity mode with auto scaling
AnswerC

On-demand tables have default throughput limits that can be increased.

Why this answer

On-demand capacity mode in DynamoDB has a default throughput quota (typically 40,000 read/write request units per second per table, though this can vary by region and account). When traffic exceeds this soft limit, DynamoDB returns ProvisionedThroughputExceededException. Requesting a service quota increase raises this ceiling, allowing the table to handle higher bursts without throttling, and requires no architectural changes or capacity management—minimizing operational overhead.

Exam trap

The trap here is that candidates assume on-demand capacity is unlimited and never throttles, but AWS imposes a default throughput quota per table that must be explicitly raised for sustained high-traffic workloads.

How to eliminate wrong answers

Option A is wrong because DAX is an in-memory cache that reduces read latency and offloads read traffic, but it does not increase the table's write throughput quota; throttling on writes or high-volume reads that bypass DAX would still occur. Option B is wrong because application-level sharding across multiple tables adds significant operational complexity (routing logic, cross-table consistency, management overhead) and is unnecessary when a simple quota increase can resolve the throttling. Option D is wrong because switching to provisioned capacity with auto scaling introduces capacity planning and scaling lag, increasing operational overhead compared to simply raising the on-demand quota; on-demand already scales instantly within its quota limits.

1449
MCQhard

A company runs a customer-facing application on Amazon RDS for MySQL. The application experiences frequent read replicas lagging behind the primary due to long-running analytics queries. The analytics team runs complex SELECT queries that scan large tables. Which design change would minimize replica lag without affecting production writes?

A.Use Amazon DynamoDB Accelerator (DAX) for caching.
B.Increase the instance size of the primary and all read replicas.
C.Enable Multi-AZ on the primary instance.
D.Create a cross-Region read replica for analytics queries.
AnswerD

Offloads analytics to a separate replica, reducing lag.

Why this answer

Creating a cross-Region read replica for analytics queries offloads the long-running SELECT statements to a separate read replica in a different AWS Region, isolating the analytics workload from the primary instance and its in-Region replicas. This prevents the analytics queries from competing for I/O and CPU resources on the primary or its local replicas, thereby minimizing replica lag without affecting production writes. Cross-Region replicas use asynchronous replication, so they can handle heavy read traffic without impacting the primary's write performance.

Exam trap

The trap here is that candidates often assume increasing instance size (Option B) or enabling Multi-AZ (Option C) will solve replica lag, but they fail to recognize that the lag is caused by resource contention from analytics queries on the same replicas, not by insufficient hardware or lack of high availability.

How to eliminate wrong answers

Option A is wrong because Amazon DynamoDB Accelerator (DAX) is an in-memory cache for DynamoDB, not for RDS for MySQL, and it does not address replica lag caused by long-running analytics queries on RDS. Option B is wrong because increasing the instance size of the primary and all read replicas may improve performance but does not isolate the analytics workload; the long-running queries on the replicas will still consume resources and cause lag, and it does not prevent the analytics queries from affecting the primary's write performance. Option C is wrong because enabling Multi-AZ on the primary instance provides high availability with a standby replica that cannot be used for reads (it is not a read replica), so it does not offload analytics queries or reduce replica lag.

1450
MCQmedium

A company uses Amazon RDS for PostgreSQL for its e-commerce platform. The application team reports increasing read latency on the primary instance during sales events. Which action should be taken to reduce read load on the primary?

A.Enable Multi-AZ deployment
B.Migrate to Amazon DynamoDB
C.Create one or more read replicas
D.Increase the instance size of the primary
AnswerC

Read replicas offload read queries from primary.

Why this answer

Creating one or more read replicas offloads read traffic from the primary RDS for PostgreSQL instance, directly addressing the increased read latency during sales events. Read replicas are asynchronous replicas that can serve SELECT queries, reducing the load on the primary without requiring application changes to the write path. This is the standard AWS solution for scaling read-heavy workloads in RDS.

Exam trap

The trap here is that candidates often confuse Multi-AZ with read replicas, assuming the standby in a Multi-AZ deployment can serve reads, but in RDS for PostgreSQL the standby is not accessible for read traffic—only Oracle and SQL Server Multi-AZ deployments offer readable standbys under specific configurations.

How to eliminate wrong answers

Option A is wrong because Multi-AZ deployment provides high availability and automatic failover via synchronous standby replication, but it does not offload read traffic—the standby is not accessible for reads in RDS for PostgreSQL. Option B is wrong because migrating to DynamoDB is a complete architectural change that is unnecessary for simply reducing read load on an existing PostgreSQL database; it would require rewriting application queries and data modeling, and it does not address the immediate symptom of read latency on the primary. Option D is wrong because increasing the instance size of the primary only vertically scales the server, which can help but is less cost-effective and does not distribute read load; it also does not leverage the horizontal read scaling that read replicas provide.

1451
MCQeasy

A company wants to migrate a 10 GB PostgreSQL database from an on-premises server to Amazon RDS for PostgreSQL. The migration can tolerate several hours of downtime. Which migration method is the MOST straightforward?

A.Use AWS Database Migration Service (AWS DMS) with ongoing replication
B.Use AWS Schema Conversion Tool (AWS SCT) to migrate the data
C.Create an Amazon RDS Read Replica from the on-premises database
D.Use pg_dump to export the database and pg_restore to import into RDS
AnswerD

This is the simplest method for a one-time migration with downtime tolerance.

Why this answer

Pg_dump and pg_restore are native PostgreSQL utilities that provide a straightforward, reliable method for migrating a 10 GB database with acceptable downtime. The 10 GB size is well within the practical limits of a logical dump, and the process requires no additional AWS services or complex configuration, making it the simplest approach for a migration that can tolerate several hours of downtime.

Exam trap

The trap here is that candidates may over-engineer the solution by choosing AWS DMS with ongoing replication, assuming it is always the best migration tool, when the question explicitly states that several hours of downtime are acceptable, making the simpler native pg_dump/pg_restore approach the most straightforward.

How to eliminate wrong answers

Option A is wrong because AWS DMS with ongoing replication is designed for minimal-downtime migrations and introduces unnecessary complexity (e.g., setting up a replication instance, source/target endpoints, and change data capture) for a scenario where several hours of downtime are acceptable. Option B is wrong because AWS SCT is used for schema conversion when migrating between different database engines (e.g., Oracle to PostgreSQL), not for migrating data between two PostgreSQL databases where the schema is already compatible. Option C is wrong because an Amazon RDS Read Replica cannot be created from an on-premises database; read replicas in RDS are only supported between RDS instances or from an RDS instance to an external source, not the reverse.

1452
MCQhard

A financial company uses Amazon DynamoDB to store customer transaction data. The compliance team requires that all data be encrypted at rest using a customer-managed AWS KMS key. Additionally, they need to ensure that the key is used only for DynamoDB and no other AWS service. How can the company meet these requirements?

A.Use a KMS key with no key policy, and rely on IAM policies to restrict access to only DynamoDB.
B.Use an AWS Organizations service control policy (SCP) to deny all AWS services except DynamoDB from using the KMS key.
C.Create a KMS key with a key policy that includes a condition such as "kms:ViaService": "dynamodb.amazonaws.com" to restrict usage to DynamoDB.
D.Create a KMS key with a key policy that allows DynamoDB to use the key, and attach an IAM policy to deny all other services.
AnswerC

This condition ensures the key can only be used through DynamoDB, preventing other services.

Why this answer

A KMS key policy can use the 'kms:ViaService' condition key to restrict usage of the key to requests that originate from DynamoDB (dynamodb.amazonaws.com). This ensures the key is used only for DynamoDB and no other AWS service. Option A is incorrect because IAM policies alone cannot restrict key usage if the key policy allows all principals; the key policy must explicitly enforce the restriction.

Option B is incorrect because AWS Organizations SCPs do not control KMS key permissions; they control permissions for IAM entities. Option D is incorrect because IAM policies cannot prevent other services from using the key if the key policy allows them; the key policy itself must include the restriction.

1453
MCQeasy

A company is using Amazon RDS for PostgreSQL to store application data. The security team wants to ensure that database audit logs are stored securely and cannot be modified after creation. Which AWS service should be used to meet this requirement?

A.AWS Key Management Service (KMS)
B.AWS CloudTrail
C.Amazon S3
D.Amazon CloudWatch Logs
AnswerD

CloudWatch Logs can store audit logs with encryption and access controls.

Why this answer

Amazon RDS for PostgreSQL can publish database audit logs to Amazon CloudWatch Logs. CloudWatch Logs provides immutability through IAM policies that prevent log modification and deletion, combined with log group encryption using AWS KMS. Option D is correct.

Option A (AWS KMS) is wrong because KMS manages encryption keys, not log storage. Option B (AWS CloudTrail) is wrong because CloudTrail records API activity, not database audit logs. Option C (Amazon S3) can store logs, but enabling immutability requires additional configuration like S3 Object Lock, making CloudWatch Logs the more straightforward and recommended service for this requirement.

1454
MCQhard

A company is migrating a 5 TB SQL Server database to Amazon RDS for SQL Server using AWS DMS. They are using Full LOB mode. The migration is failing with an error about memory allocation. What should they do to resolve this?

A.Decrease the MaxLobSize setting.
B.Increase the DMS replication instance class.
C.Split the migration into multiple DMS tasks.
D.Enable BatchApply on the DMS task.
AnswerB

More memory resolves allocation errors.

Why this answer

The memory allocation error in AWS DMS Full LOB mode occurs when the replication instance lacks sufficient memory to buffer LOB data. Increasing the replication instance class (option B) directly provides more memory, resolving the issue. Decreasing MaxLobSize (A) would truncate data, not fix allocation.

Splitting the task (C) does not increase memory per instance. Enabling BatchApply (D) improves commit efficiency but does not address memory allocation for LOBs. Therefore, B is correct.

Exam trap

The trap here is that candidates might confuse memory allocation errors with network or throughput issues and incorrectly choose to split the task or adjust LOB settings, rather than recognizing that the root cause is insufficient instance memory for Full LOB mode.

How to eliminate wrong answers

Option A is wrong because decreasing MaxLobSize would truncate LOB data, potentially causing data loss or migration failure if LOBs exceed the reduced limit; Full LOB mode is designed to handle LOBs of any size without truncation. Option C is wrong because splitting the migration into multiple tasks does not address the underlying memory constraint on the replication instance; each task still runs on the same instance and would encounter the same memory limitation. Option D is wrong because BatchApply optimizes target commit behavior for high-volume transactions but does not affect the memory allocation for LOB processing during the read phase; it is irrelevant to the memory error.

1455
MCQeasy

A company wants to automate backups for an Amazon RDS for PostgreSQL DB instance. The backup retention period should be 35 days. Which step is required?

A.Enable automated backups with a retention period of 35 days.
B.Use AWS Backup to schedule backups.
C.Create a manual snapshot every day and delete after 35 days.
D.Set up a cross-region snapshot copy.
AnswerA

RDS automated backups can be configured with a retention period up to 35 days.

Why this answer

Amazon RDS for PostgreSQL supports automated backups with a configurable retention period of up to 35 days. By enabling automated backups and setting the retention period to 35 days, the company meets the requirement without additional tooling or manual effort. Automated backups include transaction logs for point-in-time recovery, which is not available with manual snapshots alone.

Exam trap

The trap here is that candidates may overcomplicate the solution by choosing AWS Backup or manual snapshots, when the simplest and most direct method is to enable automated backups with the desired retention period, which is a native RDS feature.

How to eliminate wrong answers

Option B is wrong because AWS Backup can be used to schedule backups, but it is not required; RDS native automated backups already support a 35-day retention period natively, and AWS Backup adds no benefit for this specific requirement. Option C is wrong because creating a manual snapshot every day and deleting after 35 days is operationally complex and does not provide automated point-in-time recovery, which is a key feature of automated backups. Option D is wrong because cross-region snapshot copy is a separate feature for disaster recovery, not a step required to set the backup retention period to 35 days.

1456
MCQeasy

A company is using Amazon RDS for PostgreSQL. The security team wants to ensure that all connections to the database are encrypted in transit. Currently, applications connect using the PostgreSQL native encryption (SSL/TLS). What is the MOST secure way to enforce encrypted connections?

A.Configure the security group to only allow traffic on port 5432 from trusted IP addresses.
B.Enable the 'rds.force_ssl' parameter in the DB parameter group and restart the instance.
C.Use a custom database port that is not commonly used, such as 5433, to avoid unencrypted traffic.
D.Modify the DB parameter group to set 'ssl' to 'on' and 'require_ssl' to 'on', then reboot the instance.
AnswerB

Correct. Setting 'rds.force_ssl' (commonly referred to as 'force_ssl') to 1 in the DB parameter group and rebooting enforces SSL connections.

Why this answer

In Amazon RDS for PostgreSQL, the parameter to enforce SSL is 'rds.force_ssl' (often abbreviated as 'force_ssl'). Setting this parameter to 1 in the DB parameter group and rebooting the instance forces all connections to use SSL/TLS. Option A is incorrect because security groups control network access but do not enforce encryption.

Option C is incorrect because changing the port only obscures the port number, not enforce encryption. Option D is incorrect because 'require_ssl' is not a valid parameter in RDS PostgreSQL; the correct parameter is 'rds.force_ssl'.

Exam trap

Candidates often confuse the parameter 'require_ssl' (common in on-premises PostgreSQL) with the RDS-specific parameter 'rds.force_ssl'.

1457
MCQhard

A company uses Amazon ElastiCache for Redis as a caching layer for a database. The cache cluster has one primary and one replica node. During a maintenance event, the primary node fails and the replica is promoted. After promotion, the application experiences increased latency. What should a database specialist do to reduce the impact of future failovers?

A.Disable replica reads to avoid stale data.
B.Enable automatic backups with a retention period of 35 days.
C.Enable cluster mode and distribute the cache across multiple shards.
D.Add more replica nodes to the cluster.
AnswerC

Cluster mode spreads data across shards, reducing the impact of a single node failure.

Why this answer

Enabling cluster mode on the ElastiCache for Redis cluster distributes data across multiple shards, each with its own primary and replica nodes. During a failover, only the data in the failed shard's primary needs to be rebuilt, limiting the cache miss impact. Adding more replicas (Option D) does not reduce the latency impact because a single primary still serves all requests after failover, and the new primary must warm its cache from scratch.

Disabling replica reads (Option A) would not address the latency increase. Automatic backups (Option B) do not affect runtime performance after failover.

1458
MCQeasy

An application is experiencing increased latency when writing to an Amazon DynamoDB table. The table uses on-demand capacity mode. The CloudWatch metric 'WriteThrottleEvents' is zero. What is the most likely cause of the increased latency?

A.The write capacity units (WCUs) are set too low.
B.DynamoDB Accelerator (DAX) is not configured for writes.
C.A hot partition is causing excessive write traffic to a single partition.
D.The table is experiencing write throttling due to exceeding the provisioned write capacity.
AnswerC

Hot partitions can cause increased latency even when overall throughput is within limits, as a single partition's capacity is constrained.

Why this answer

On-demand DynamoDB tables can experience throttling if you exceed the previous peak traffic by more than double in a short time, but since WriteThrottleEvents is zero, the latency is likely due to a hot partition causing uneven traffic distribution. Option A is incorrect because on-demand capacity does not use WCUs; throughput is automatically scaled. Option B is incorrect because DynamoDB Accelerator (DAX) is a read cache and does not affect write latency.

Option D is incorrect because on-demand tables do not have provisioned capacity; throttling would be indicated by WriteThrottleEvents, which is zero.

1459
MCQeasy

Refer to the exhibit. A developer created a DynamoDB table 'UserSessions' with a simple primary key. The application needs to query by user_id as well. What design change should the developer make to support this query efficiently?

A.Use a Scan operation with a filter
B.Add a sort key to the table
C.Create a Local Secondary Index on user_id
D.Create a Global Secondary Index on user_id
AnswerD

A GSI enables efficient querying on user_id.

Why this answer

A Global Secondary Index (GSI) on user_id allows efficient querying by user_id without altering the base table's primary key structure. The base table uses a simple primary key (likely session_id), and a GSI provides a separate index with its own partition key (user_id) to support non-key attribute queries with eventual consistency, enabling the application to query by user_id efficiently without scanning the entire table.

Exam trap

AWS often tests the misconception that a Local Secondary Index can be used to query on any attribute, but the trap here is that an LSI requires the same partition key as the base table, so it cannot index user_id as a partition key unless user_id is already the base table's partition key.

How to eliminate wrong answers

Option A is wrong because a Scan operation with a filter reads every item in the table, incurring high read capacity consumption and latency, which is inefficient for frequent queries. Option B is wrong because adding a sort key to the table would change the primary key structure, requiring a new table or migration, and does not directly support querying by user_id unless user_id is already the partition key. Option C is wrong because a Local Secondary Index (LSI) can only be created at table creation time and shares the same partition key as the base table; if the base table's partition key is not user_id, an LSI cannot index user_id as a partition key, making it unsuitable for this use case.

1460
Multi-Selecteasy

Which TWO design patterns are commonly used to handle hot partitions in Amazon DynamoDB? (Choose 2.)

Select 2 answers
A.Write sharding
B.Decreasing write capacity units
C.Using a single partition key
D.Increasing read capacity units
E.Adding random suffixes to partition keys
AnswersA, E

Distributes writes across many partition key values.

Why this answer

Write sharding distributes writes across multiple partition keys to prevent a single partition from exceeding the 1,000 WCU limit. Adding random suffixes to partition keys is a specific write sharding technique that spreads writes across many partitions, avoiding hot spots.

Exam trap

AWS often tests the misconception that increasing capacity units alone resolves hot partitions, but the real solution requires redistributing the workload across partitions via sharding or suffix-based strategies.

1461
MCQhard

An e-commerce platform uses Amazon DynamoDB for a shopping cart table with partition key 'user_id' and sort key 'product_id'. The table experiences throttled write requests during flash sales. The access pattern includes reading the entire cart at checkout. Which design change would improve write performance without changing the read pattern?

A.Enable DynamoDB Accelerator (DAX) to cache writes
B.Increase the provisioned write capacity units (WCU) to a higher value
C.Change the table design to use only partition key 'user_id' and remove the sort key
D.Enable DynamoDB Adaptive Capacity and ensure the table uses on-demand capacity mode
AnswerD

Adaptive capacity helps distribute traffic across partitions, and on-demand mode handles spikes.

Why this answer

Enabling DynamoDB Adaptive Capacity with on-demand capacity mode automatically scales write capacity to handle traffic spikes during flash sales without requiring manual provisioning. This eliminates throttling while preserving the existing table schema (partition key 'user_id' and sort key 'product_id'), so the read pattern of querying the entire cart by user_id remains unchanged.

Exam trap

The trap here is that candidates often confuse DAX as a write accelerator or assume that simply increasing provisioned capacity is sufficient, overlooking that on-demand mode with adaptive capacity is the correct solution for unpredictable traffic spikes without schema changes.

How to eliminate wrong answers

Option A is wrong because DAX is an in-memory cache that accelerates reads, not writes; it cannot improve write performance or reduce write throttling. Option B is wrong because increasing provisioned WCU only helps if the traffic pattern is predictable; during flash sales, traffic spikes can still exceed the new provisioned limit, leading to continued throttling unless capacity is over-provisioned. Option C is wrong because removing the sort key 'product_id' would break the ability to store multiple products per user in the cart, fundamentally altering the data model and potentially causing data loss or overwrites.

1462
MCQhard

A company is migrating a 10 TB Oracle database to Amazon Aurora PostgreSQL. The migration uses AWS DMS with CDC. After the full load, the CDC phase is falling behind by several minutes. The source Oracle database generates 500 MB of redo logs per minute. Which action will most likely improve CDC performance?

A.Disable CDC and perform a full load only
B.Increase the number of parallel apply threads on the DMS task
C.Reduce the batch size in the DMS task settings
D.Increase the source Oracle redo log size
AnswerB

More parallel apply threads can increase throughput for applying changes to the target.

Why this answer

Increasing the number of parallel apply threads on the DMS task allows the target Aurora PostgreSQL database to apply changes concurrently, reducing the bottleneck caused by sequential apply. With 500 MB/min of redo logs, the single-threaded default apply cannot keep pace, so parallel apply directly addresses the lag by distributing the load across multiple threads.

Exam trap

The trap here is that candidates often confuse batch size with parallelism, assuming reducing batch size will speed up apply, when in fact it increases overhead and slows down CDC, while the correct solution is to increase parallelism to match the high redo log generation rate.

How to eliminate wrong answers

Option A is wrong because disabling CDC and performing only a full load would not capture ongoing changes, defeating the purpose of a migration with minimal downtime; the issue is CDC performance, not the full load. Option C is wrong because reducing the batch size in DMS task settings would decrease the number of changes applied per transaction, increasing the number of round trips and likely worsening the lag, not improving it. Option D is wrong because increasing the source Oracle redo log size does not affect DMS CDC performance; DMS reads redo logs at the rate they are generated, and larger logs do not change the volume of data or the apply speed on the target.

1463
MCQmedium

A company is using Amazon Neptune to run graph queries. The cluster has one writer and two reader instances. After a major version upgrade, the query performance degrades for complex traversal queries. The database specialist suspects that the query optimizer is not using indexes effectively. Which action should the specialist take to identify the issue?

A.Drop and recreate all indexes to ensure they are up to date.
B.Enable the Neptune explain plan feature and review the output.
C.Add more reader instances to distribute the query load.
D.Check the CloudWatch metrics for CPU utilization and query latency.
AnswerB

Explain plan shows query execution steps and index usage.

Why this answer

Enabling the Neptune explain plan feature provides detailed information about how queries are executed, including index usage. This helps identify if the query optimizer is failing to use indexes effectively after the upgrade. Option A is incorrect because dropping and recreating indexes is unnecessary and disruptive; indexes are automatically maintained.

Option C is incorrect because adding reader instances distributes read load but does not fix query optimization issues. Option D is incorrect because CloudWatch metrics show system-level performance, not query execution details.

Exam trap

The trap is that adding readers or checking metrics might seem helpful for performance, but the question specifically asks to identify index usage—only the explain plan provides that insight.

1464
Multi-Selectmedium

Which TWO factors should you consider when choosing between Amazon RDS and Amazon DynamoDB for a new application?

Select 2 answers
A.RDS requires a predefined schema, while DynamoDB is schema-less.
B.DynamoDB can only be accessed from within a VPC, while RDS can be public.
C.Only RDS supports Multi-AZ deployments for high availability.
D.Both services support encryption at rest and in transit.
E.DynamoDB is better suited for unstructured data, while RDS is better for structured data with complex relationships.
AnswersA, E

This is a key difference that affects application design.

Why this answer

Amazon RDS (relational database service) requires a predefined schema with tables, columns, and data types before data can be inserted, enforcing ACID compliance and referential integrity. In contrast, Amazon DynamoDB is a NoSQL key-value and document database that is schema-less, allowing you to store items with varying attributes without upfront schema definition, which is ideal for agile development and unstructured data.

Exam trap

The trap here is that candidates often assume DynamoDB cannot be accessed publicly or that only RDS supports Multi-AZ, but in reality both services offer these features, and the key differentiator is the data model (schema vs. schema-less) and the nature of the data (structured with relationships vs. unstructured).

1465
MCQmedium

A company is migrating a 2 TB Oracle database from on-premises to Amazon RDS for Oracle using AWS DMS. The migration completes successfully, but the new RDS instance shows higher CPU utilization than the on-premises database for the same workload. Which action is MOST likely to reduce CPU utilization on RDS?

A.Disable automated backups to free resources
B.Enable Oracle Multitenant architecture
C.Increase the allocated storage to improve IOPS
D.Change the RDS instance to a burstable class
AnswerC

Increasing allocated storage improves IOPS, which reduces I/O wait times and can significantly lower CPU utilization if the database is I/O-bound. This is often the most straightforward first step.

Why this answer

Increasing allocated storage on Amazon RDS for Oracle directly improves the provisioned IOPS (if using Provisioned IOPS) or baseline IOPS (for gp2/gp3). Higher IOPS reduces I/O wait time, which is a common cause of high CPU utilization when the database is I/O-bound. Query optimization or instance resizing may also help, but increasing storage is a simple, low-risk action that often alleviates I/O-related CPU pressure.

Option B is incorrect because Oracle Multitenant (CDB/PDB) architecture adds container management overhead and is not designed to reduce CPU utilization; also, converting an existing non-CDB RDS instance to CDB is not supported.

Exam trap

The trap is assuming that CPU utilization is always due to compute limits and overlooking the impact of I/O wait. Candidates may also incorrectly believe Oracle Multitenant reduces CPU, but it actually adds overhead and is not applicable to existing RDS instances.

How to eliminate wrong answers

Option A is wrong because disabling automated backups does not significantly reduce CPU utilization; backups primarily affect I/O and storage, not CPU, and disabling them would compromise recovery capabilities without addressing the root cause. Option C is wrong because increasing allocated storage to improve IOPS addresses I/O bottlenecks, not CPU utilization; higher IOPS can reduce wait times but does not lower CPU usage for the same workload. Option D is wrong because changing to a burstable class (e.g., db.t3) may reduce cost but can actually increase CPU utilization due to CPU credits and throttling under sustained load, making it unsuitable for reducing CPU usage.

1466
MCQhard

A company is using Amazon DynamoDB to store financial transactions. The security team requires that all access to the table be logged for auditing, and that any unauthorized access attempts trigger an immediate alert. The company has enabled AWS CloudTrail to log all DynamoDB API calls. However, the security team is concerned that CloudTrail logs may not capture all access patterns, such as queries that return no results. Which additional step should the company take to ensure comprehensive auditing and alerting?

A.Configure Amazon Inspector to assess the DynamoDB table for vulnerabilities.
B.Enable DynamoDB Accelerator (DAX) and configure it to log all read requests.
C.Create a CloudWatch Logs metric filter on the CloudTrail log group to detect unauthorized access attempts and set up a CloudWatch alarm.
D.Enable VPC Flow Logs on the subnet where DynamoDB endpoints are deployed.
AnswerC

CloudWatch Logs can analyze CloudTrail logs and trigger alarms based on patterns.

Why this answer

CloudWatch Logs can be used to monitor CloudTrail logs and trigger alerts on specific patterns like unauthorized access. CloudTrail logs all DynamoDB API calls, including queries that return no results. By creating a metric filter on the CloudTrail log group for unauthorized access patterns (e.g., AccessDenied exceptions) and setting a CloudWatch alarm, the company can get immediate alerts.

Option A (Amazon Inspector) is for vulnerability assessment, not access logging. Option B (DAX) is a caching layer and does not log all read requests; it only caches and can be configured for logging, but it does not replace CloudTrail for auditing. Option D (VPC Flow Logs) captures network traffic, not API call details.

1467
MCQhard

An IAM policy is attached to a user. What is the effect of this policy on the user's ability to delete the DB instance named prod-db? The policy is: ```json { "Version": "2012-10-17", "Statement": [ { "Effect": "Deny", "Action": "rds:DeleteDBInstance", "Resource": "arn:aws:rds:us-east-1:123456789012:db:prod-db" }, { "Effect": "Allow", "Action": "rds:*", "Resource": "*" } ] } ```

A.The user can delete the DB instance only after creating a final snapshot.
B.The user can delete the DB instance because the Allow statement grants all actions.
C.The user cannot delete the DB instance because the Deny statement explicitly denies it.
D.The user can delete the DB instance because the Allow statement is broader and applies to all resources.
AnswerC

Assumes an explicit Deny statement, but without the policy, we cannot confirm whether a Deny exists.

Why this answer

In IAM policy evaluation, an explicit Deny overrides any Allow. The policy explicitly denies the `rds:DeleteDBInstance` action on the `prod-db` resource, so the user cannot delete the DB instance, regardless of the Allow statement granting all RDS actions.

1468
MCQeasy

A startup is building a social media application with a news feed feature. The feed must be personalized and updated in real-time as users post. Which AWS database service is best suited for this workload?

A.Amazon DynamoDB with Global Secondary Indexes
B.Amazon S3 with Select and Glacier
C.Amazon RDS for PostgreSQL with read replicas
D.Amazon ElastiCache for Redis with sorted sets and pub/sub
AnswerD

Redis provides real-time data structures and pub/sub for feeds.

Why this answer

Amazon ElastiCache for Redis is the best choice because it provides in-memory data structures like sorted sets for ranking and scoring personalized feeds, and pub/sub for real-time notifications when new posts are published. This combination enables low-latency, real-time feed updates without the overhead of disk-based storage, making it ideal for a social media news feed that must be both personalized and updated in real-time.

Exam trap

The trap here is that candidates often choose DynamoDB or RDS because they are familiar with them for data storage, but they overlook the need for real-time, in-memory operations and the specific data structures (sorted sets, pub/sub) that only ElastiCache for Redis provides for this workload.

How to eliminate wrong answers

Option A is wrong because DynamoDB with Global Secondary Indexes is a NoSQL database optimized for key-value and document workloads, but it lacks native pub/sub and sorted set capabilities required for real-time feed personalization and push updates. Option B is wrong because S3 is an object storage service designed for static data archiving and retrieval, not for low-latency, real-time read/write operations; S3 Select and Glacier are for querying and cold storage, respectively, and cannot support live feed updates. Option C is wrong because RDS for PostgreSQL with read replicas is a relational database that can handle complex queries but introduces higher latency for real-time updates and lacks built-in sorted sets and pub/sub, making it unsuitable for high-throughput, low-latency feed personalization.

1469
MCQhard

A company uses Amazon DynamoDB as the primary database for a global gaming application. The application requires single-digit millisecond latency for user profile lookups by user ID. However, some queries need to retrieve all active users in a region (e.g., 'us-east-1') for administrative dashboards, and these queries currently perform full table scans, causing high costs and throttling. What design approach should be taken to optimize this?

A.Implement DynamoDB Accelerator (DAX) to cache the dashboard queries.
B.Increase the read capacity units (RCUs) on the base table.
C.Create a global secondary index (GSI) on the region attribute.
D.Create a local secondary index (LSI) on the region attribute.
AnswerC

A GSI allows efficient querying on region without scanning the base table, reducing cost and throttling.

Why this answer

Creating a Global Secondary Index (GSI) on the 'region' attribute allows the administrative dashboard queries to retrieve all active users in a specific region using an efficient index scan instead of a full table scan. This reduces read capacity consumption, avoids throttling, and maintains single-digit millisecond latency for the indexed queries, while the base table remains optimized for user ID lookups.

Exam trap

The trap here is that candidates often confuse LSIs with GSIs, assuming an LSI can be used to query by a non-key attribute like region, but LSIs are limited to the same partition key as the base table and cannot avoid a full scan when the query predicate is on a different partition key.

How to eliminate wrong answers

Option A is wrong because DynamoDB Accelerator (DAX) is an in-memory cache that accelerates reads on the base table or existing indexes, but it does not eliminate the need for a full table scan when querying by region; it would only cache the results of expensive scans, not prevent them. Option B is wrong because increasing read capacity units (RCUs) on the base table would temporarily reduce throttling but does not address the root cause—full table scans are inherently inefficient and costly at scale, and higher RCUs only mask the problem while increasing costs. Option D is wrong because a Local Secondary Index (LSI) can only be created at table creation time and shares the same partition key as the base table; since the base table's partition key is user ID (not region), an LSI on region would still require a full scan across all partitions to retrieve all users in a region, providing no performance benefit.

1470
MCQmedium

A company is migrating its on-premises Oracle database to Amazon RDS for Oracle. As part of the migration, they need to ensure that all sensitive data is encrypted at rest using AWS Key Management Service (AWS KMS). Which configuration step is required to achieve this?

A.Modify the existing DB instance to enable encryption.
B.Configure SSL/TLS on the DB instance to encrypt data at rest.
C.Use AWS CloudHSM to generate and store the encryption keys.
D.Create a new encrypted DB instance by enabling encryption and specifying a KMS key.
AnswerD

Creating a new encrypted RDS instance with a KMS key is the correct way to enable encryption at rest.

Why this answer

To encrypt an Amazon RDS for Oracle DB instance at rest using AWS KMS, you must enable encryption when creating the DB instance and specify a KMS key. Encryption cannot be added to an existing unencrypted DB instance directly; you must take a snapshot, create an encrypted copy, and restore it. Option A is incorrect because modifying an existing instance does not allow enabling encryption.

Option B is incorrect because SSL/TLS encrypts data in transit, not at rest. Option C is incorrect because while CloudHSM can be used, AWS KMS is the simpler and more common approach, and the question specifically mentions AWS KMS.

1471
MCQhard

A company uses Amazon DynamoDB for a gaming leaderboard. The application updates scores frequently. Reads must be strongly consistent, and writes must be optimized for cost. Which table design minimizes cost while meeting consistency requirements?

A.Use Amazon DynamoDB Accelerator (DAX) for caching.
B.Use eventually consistent reads with a conditional write.
C.Store scores in Amazon S3 and use S3 Select for reads.
D.Use DynamoDB Streams to replicate reads to a separate table.
AnswerA

DAX provides in-memory caching with strong consistency, reducing RCU cost.

Why this answer

Amazon DynamoDB Accelerator (DAX) provides an in-memory cache that supports strongly consistent reads, which meets the application's requirement for strongly consistent reads. By caching frequently accessed leaderboard data, DAX reduces the number of read capacity units consumed from the DynamoDB table, thereby lowering read costs. Writes are still performed directly on the DynamoDB table, and DAX does not affect write costs, so the design optimizes overall cost while maintaining consistency.

Exam trap

The trap here is that candidates may assume that eventually consistent reads are sufficient for a leaderboard, or that caching with DAX is only for performance and not for cost optimization, but the question explicitly requires strongly consistent reads and cost minimization, making DAX the correct choice.

How to eliminate wrong answers

Option B is wrong because eventually consistent reads do not meet the requirement for strongly consistent reads, and conditional writes are used for optimistic locking, not for consistency or cost optimization. Option C is wrong because storing scores in Amazon S3 and using S3 Select for reads introduces significant latency and does not support the low-latency, high-frequency updates required for a gaming leaderboard; S3 is not designed for real-time strongly consistent reads. Option D is wrong because using DynamoDB Streams to replicate reads to a separate table adds complexity, latency, and additional storage costs without providing strongly consistent reads from the replica; DynamoDB Streams is for change data capture, not for read consistency.

1472
Multi-Selecteasy

A developer is building a serverless application that uses Amazon DynamoDB. The application needs to access the database from an AWS Lambda function. The security team mandates that the Lambda function should not use long-term AWS credentials. Which TWO steps should be taken to securely grant access? (Choose TWO.)

Select 2 answers
A.Hardcode the AWS access key ID and secret access key in the Lambda environment variables.
B.Create an IAM role with a policy that allows DynamoDB actions.
C.Configure the Lambda function to access the internet for authentication.
D.Store the database credentials in AWS Secrets Manager and retrieve them in the Lambda function.
E.Attach the IAM role to the Lambda function's execution role.
AnswersB, E

IAM role provides temporary credentials.

Why this answer

To securely grant an AWS Lambda function access to DynamoDB without using long-term credentials, you should create an IAM role that grants the necessary DynamoDB permissions (option B) and then attach that IAM role to the Lambda function's execution role (option E). This allows Lambda to obtain temporary credentials via the IAM role, eliminating the need for hardcoded credentials. Option A is incorrect because hardcoding credentials in environment variables is not secure and uses long-term credentials.

Option C is incorrect because the function does not need internet access for authentication; IAM roles provide temporary credentials internally. Option D is incorrect because while Secrets Manager can store credentials, it still requires managing secrets and may not align with the mandate to avoid long-term credentials; IAM roles are the preferred method.

1473
MCQeasy

A database administrator notices that Amazon RDS for MySQL is experiencing high CPU utilization during peak hours. The application is read-heavy with many SELECT queries. Which action is most cost-effective to improve performance?

A.Create one or more read replicas and direct read traffic to them.
B.Increase the DB instance class to a larger size.
C.Increase the allocated storage to improve I/O throughput.
D.Scale up the DB instance vertically to a higher vCPU count.
AnswerA

Read replicas distribute read workload, reducing primary CPU.

Why this answer

The most cost-effective solution for a read-heavy, high-CPU workload on Amazon RDS for MySQL is to create one or more read replicas and direct read traffic to them. This offloads SELECT queries from the primary DB instance, reducing CPU utilization without incurring the high cost of upgrading the primary instance. Option B (increasing DB instance class) and Option D (vertical scaling to higher vCPU) are more expensive and not cost-effective.

Option C (increasing allocated storage) improves I/O throughput but does not directly reduce CPU utilization.

1474
MCQmedium

A company is migrating an on-premises PostgreSQL database to Amazon RDS for PostgreSQL. The database is 2 TB in size and has a high write volume. The migration window is limited to 4 hours. Which migration approach provides the fastest initial load with minimal downtime?

A.Use pg_dump to export the database and pg_restore to import into RDS.
B.Create a read replica of the on-premises database and promote it to a standalone database in RDS.
C.Use AWS Schema Conversion Tool (AWS SCT) to convert the schema and then use AWS DMS for ongoing replication.
D.Use AWS DMS with PostgreSQL as source and target, using native logical replication for full load and ongoing replication.
AnswerD

DMS with logical replication can perform a full load quickly and then keep the target in sync with minimal downtime.

Why this answer

AWS DMS with native logical replication (using PostgreSQL's built-in logical replication slots) performs a full load of the 2 TB database while simultaneously capturing ongoing changes, enabling near-zero downtime during the migration window. This approach is optimized for high-write volumes and limited windows, as it avoids the need for a separate export/import step and minimizes the time the source database is unavailable.

Exam trap

The trap here is that candidates often confuse AWS DMS with schema conversion tools (AWS SCT) or assume that traditional dump/restore tools are sufficient for large, high-write databases, overlooking DMS's native logical replication capability for same-engine migrations with minimal downtime.

How to eliminate wrong answers

Option A is wrong because pg_dump/pg_restore requires a full export and import of the 2 TB database, which typically takes longer than 4 hours for high-write workloads and incurs significant downtime during the final cutover. Option B is wrong because creating a read replica of an on-premises PostgreSQL database is not supported; read replicas are a feature of Amazon RDS, not on-premises instances, and promoting a replica to a standalone database does not apply to cross-environment migrations. Option C is wrong because AWS SCT is used for schema conversion when migrating between different database engines (e.g., Oracle to PostgreSQL), not for same-engine PostgreSQL migrations, and using DMS with SCT adds unnecessary complexity without improving speed.

1475
MCQeasy

A database specialist needs to monitor the resource utilization of Amazon RDS DB instances. Which AWS service provides OS-level metrics such as memory, disk, and CPU usage?

A.Amazon RDS Performance Insights
B.Amazon RDS Enhanced Monitoring
C.Amazon CloudWatch
D.AWS CloudTrail
AnswerB

Enhanced Monitoring provides OS-level metrics.

Why this answer

Amazon RDS Enhanced Monitoring provides OS-level metrics (e.g., memory, disk, CPU usage) by running an agent on the RDS host and delivering logs to CloudWatch Logs. This is the only service that exposes hypervisor-level and guest OS metrics for RDS instances, enabling granular troubleshooting of resource contention.

Exam trap

The trap here is that candidates confuse CloudWatch’s basic RDS metrics (which are hypervisor-level) with the OS-level metrics provided only by Enhanced Monitoring, leading them to incorrectly select CloudWatch.

How to eliminate wrong answers

Option A is wrong because Amazon RDS Performance Insights focuses on database workload analysis (e.g., wait events, SQL queries) and does not provide OS-level metrics like memory or disk usage. Option C is wrong because Amazon CloudWatch provides basic RDS metrics (e.g., CPUUtilization, FreeableMemory) at the hypervisor level, but not the detailed OS-level metrics (e.g., file system disk usage, process list) that Enhanced Monitoring offers. Option D is wrong because AWS CloudTrail records API activity for governance and auditing, not real-time OS-level resource utilization.

1476
MCQhard

A database specialist is troubleshooting a performance issue on a self-managed PostgreSQL database that they plan to migrate to Amazon RDS. The database has a high number of 'idle in transaction' connections. What is the impact of these connections on the database?

A.They increase CPU usage due to constant polling.
B.They hold locks and prevent cleanup of dead tuples, leading to bloat.
C.They prevent new connections from being established.
D.They cause increased disk I/O from write-ahead logging.
AnswerB

Idle transactions keep locks and prevent autovacuum from marking dead tuples.

Why this answer

Idle-in-transaction connections hold locks and prevent PostgreSQL's autovacuum from cleaning up dead tuples, leading to table bloat and performance degradation. Option A is wrong because idle transactions do not cause constant polling; CPU usage remains low. Option C is wrong because they do not prevent new connections from being established unless the max_connections limit is reached.

Option D is wrong because idle transactions do not significantly increase write-ahead logging.

1477
Multi-Selecthard

A company is designing a document database on Amazon DocumentDB for a content management system. Which TWO design practices improve query performance and reduce costs?

Select 1 answer
A.Shard data based on access patterns to distribute load.
B.Design documents to avoid joins by frequently using $lookup.
C.Avoid denormalization to maintain strict normal forms.
D.Store all documents in a single collection without indexes to reduce overhead.
E.Use appropriate indexes to support common query patterns.
AnswersE

Correct: Using appropriate indexes minimizes the amount of data scanned, speeding up queries and reducing I/O costs.

Why this answer

In Amazon DocumentDB, only Option E (Use appropriate indexes to support common query patterns) is correct. Sharding (Option A) is not supported by DocumentDB. Options B, C, and D are incorrect because they would degrade performance or increase costs.

Note: The question asks for two, but only one option is correct.

Exam trap

Candidates often think DocumentDB supports native sharding like MongoDB. In reality, Amazon DocumentDB does not support sharding. Proper indexing is the primary performance optimization for DocumentDB.

1478
MCQhard

A company is using Amazon Redshift for data warehousing. The data engineering team notices that queries are taking longer than expected. The cluster has two nodes of type dc2.large. The database specialist checks the system tables and finds that many queries are using the disk for temporary storage. Which action should the specialist take to improve query performance?

A.Add distribution keys to the tables to improve data distribution.
B.Enable concurrency scaling to offload queries to additional clusters.
C.Increase the number of nodes to three to distribute the workload.
D.Upgrade the cluster to a node type with more memory, such as ra3.xlplus.
AnswerD

Upgrading to ra3.xlplus nodes increases memory per node (from ~15 GB to ~32 GB), directly reducing the likelihood of disk spill for memory-intensive queries, thus improving query performance.

Why this answer

Disk spill to temporary storage indicates insufficient memory per node. Upgrading from dc2.large (∼15 GB RAM) to ra3.xlplus (∼32 GB RAM) doubles per-node memory, reducing disk spill and improving query performance. Option A is incorrect because distribution keys improve data distribution, not memory.

Option B is incorrect because concurrency scaling manages concurrent queries, not per-query memory. Option C is incorrect because adding a node increases total cluster memory but per-node memory stays the same; disk spill occurs per node, so the problem persists.

1479
MCQeasy

A company is deploying Amazon RDS for MySQL in a Multi-AZ configuration for high availability. The database must be able to automatically failover to a standby in another Availability Zone. Which RDS feature enables this?

A.Multi-AZ deployment
B.Automated backups
C.Enhanced Monitoring
D.Read Replicas
AnswerA

Multi-AZ automatically fails over to standby in another AZ.

Why this answer

Multi-AZ deployment for Amazon RDS automatically provisions and maintains a synchronous standby replica in a different Availability Zone. When a failure is detected, Amazon RDS automatically fails over to the standby, ensuring high availability without manual intervention. This is the native RDS feature designed specifically for automatic failover across AZs.

Exam trap

The trap here is confusing Read Replicas (asynchronous, manual promotion) with Multi-AZ (synchronous, automatic failover), as both involve replicas in different AZs but serve fundamentally different purposes.

How to eliminate wrong answers

Option B is wrong because Automated backups are for point-in-time recovery and retention of database snapshots, not for automatic failover or synchronous replication. Option C is wrong because Enhanced Monitoring provides real-time OS-level metrics for performance analysis, not failover capabilities. Option D is wrong because Read Replicas are asynchronous replicas used for read scaling and can be promoted manually, but they do not provide automatic failover or synchronous replication for high availability.

1480
Multi-Selecthard

A company is migrating its on-premises Oracle database to Amazon RDS for Oracle. The database specialist needs to monitor the migration process and ensure data consistency. Which TWO AWS services should be used together to continuously monitor the replication lag and data integrity?

Select 2 answers
A.Amazon RDS Performance Insights
B.AWS Schema Conversion Tool (AWS SCT)
C.AWS Database Migration Service (AWS DMS)
D.AWS Glue
E.Amazon CloudWatch
AnswersC, E

DMS provides replication tasks and publishes latency metrics.

Why this answer

AWS DMS is the correct service because it is specifically designed for database migrations and provides built-in monitoring of replication lag via the 'CDC latency' metric. Amazon CloudWatch is the correct complementary service because it collects and visualizes DMS metrics, including replication lag and task status, and can trigger alarms if data integrity or latency thresholds are breached.

Exam trap

The trap here is that candidates may confuse AWS DMS with AWS Glue or SCT, thinking those services also handle continuous replication monitoring, but only DMS provides CDC metrics that CloudWatch can monitor for lag and integrity.

1481
Drag & Dropmedium

Arrange the steps to switch over from a primary Amazon RDS for Oracle DB instance to a standby in a Multi-AZ deployment (planned failover) in the correct order.

Drag steps to the numbered slots on the right, or tap a step then tap a slot.

Steps
Order
1Step 1
2Step 2
3Step 3
4Step 4

Why this order

Planned failover involves verifying standby health, initiating failover, and confirming connectivity.

1482
MCQeasy

A startup is running an Amazon RDS for PostgreSQL database for its web application. The database size is 50 GB. The company wants to implement a backup strategy that allows point-in-time recovery (PITR) to any point within the last 35 days with minimal storage cost. Which backup strategy should be used?

A.Enable automated backups with a retention period of 35 days.
B.Use AWS Backup to schedule daily snapshots with 35-day retention.
C.Enable automated backups with 7-day retention and copy snapshots to another region.
D.Take manual snapshots daily and retain 35 snapshots.
AnswerA

Automated backups support PITR up to 35 days.

Why this answer

Automated backups in Amazon RDS provide point-in-time recovery (PITR) within the retention period, and the maximum retention period is 35 days. Enabling automated backups with a 35-day retention period meets the requirement for PITR to any point within the last 35 days with minimal storage cost, as the backup storage is charged only for the incremental changes. Option B (AWS Backup) introduces additional cost and complexity without adding value for this scenario.

Option C (7-day retention with cross-region copy) does not achieve 35-day PITR and incurs cross-region transfer costs. Option D (daily manual snapshots) does not provide PITR and results in higher storage costs due to full snapshots.

1483
Multi-Selecthard

A company uses Amazon DynamoDB with provisioned capacity for a critical application. During a traffic spike, the table experiences throttling on write requests. The DBA needs to resolve the issue quickly. Which THREE actions should the DBA take? (Choose THREE.)

Select 3 answers
A.Switch the table to on-demand capacity mode.
B.Implement exponential backoff in the application's write requests.
C.Temporarily increase the write capacity units (WCU) using the AWS Console or CLI.
D.Enable auto scaling for the table to automatically adjust capacity.
E.Enable DynamoDB Accelerator (DAX) to cache writes.
AnswersB, C, D

Exponential backoff reduces retry collisions.

Why this answer

Uses exponential backoff to handle throttling by retrying requests with increasing delays. Option C temporarily increases write capacity units (WCU) to handle the spike. Option D enables auto scaling to automatically adjust capacity.

Option A is incorrect because switching to on-demand may resolve throttling but is not one of the three correct actions; the question asks for three specific actions, and B, C, D are the correct combination. Option E is wrong because DAX is a caching layer for reads, not writes, and does not resolve write throttling.

1484
MCQhard

A company uses Amazon DynamoDB for a real-time analytics platform. The table has a partition key of 'customer_id' and a sort key of 'event_timestamp'. The table receives 50,000 write requests per second, evenly distributed across 10,000 customers. The application frequently queries the last 10 events for a given customer. The company notices that some queries are throttled during peak hours. The table's write capacity is set to 50,000 WCUs, and read capacity to 10,000 RCUs. The throttled queries are read requests. What is the most likely cause of the throttling, and what should be done to resolve it?

A.Increase the write capacity units to handle the write load.
B.Increase the read capacity units to 20,000 RCUs.
C.Optimize the query by using Query with KeyConditionExpression on the sort key and Limit=10.
D.Add a global secondary index with the same keys to distribute read load.
AnswerC

This ensures the query reads only the necessary items, reducing RCU consumption.

Why this answer

The throttling occurs because the application uses Scan or an inefficient query pattern that consumes excessive read capacity. Using Query with KeyConditionExpression on the sort key and Limit=10 retrieves only the last 10 events per customer efficiently, reducing read consumption and avoiding throttling without increasing RCUs.

Exam trap

The DBS-C01 exam often tests the misconception that throttling always requires increasing capacity, when in fact optimizing the access pattern with Query and Limit can resolve the issue without additional cost.

How to eliminate wrong answers

Option A is wrong because the issue is read throttling, not write throttling, and write capacity is already sufficient at 50,000 WCUs. Option B is wrong because increasing RCUs to 20,000 would mask the inefficiency without addressing the root cause—poor query design that consumes more capacity than necessary. Option D is wrong because adding a GSI with the same keys would not distribute read load differently; the base table already has the required keys, and a GSI would not improve query efficiency for this access pattern.

1485
MCQmedium

A company is migrating a 1 TB SQL Server database to Amazon RDS for SQL Server. They need to minimize downtime and support ongoing replication. Which combination of AWS services should they use?

A.AWS SCT and AWS DMS
B.AWS S3 and AWS Glue
C.AWS DMS alone
D.Native SQL Server backup and restore to RDS
AnswerA

SCT converts schema, DMS migrates data with ongoing replication.

Why this answer

AWS DMS (Database Migration Service) can perform a full load of the 1 TB SQL Server database and then continuously replicate ongoing changes using change data capture (CDC) to minimize downtime. AWS SCT (Schema Conversion Tool) is used to assess and convert any incompatible schema objects (e.g., stored procedures, indexes) from the source SQL Server to the target RDS for SQL Server, ensuring a smooth migration. Together, they provide a complete solution for heterogeneous or homogeneous migration with minimal downtime and ongoing replication.

Exam trap

The trap here is that candidates often assume AWS DMS alone is sufficient for any migration, forgetting that schema conversion (via AWS SCT) is critical when moving to a different SQL Server version or when source objects are incompatible with RDS, which the exam tests as a required combination for minimizing downtime and supporting ongoing replication.

How to eliminate wrong answers

Option B is wrong because AWS S3 and AWS Glue are designed for data lakes and ETL (Extract, Transform, Load) workflows, not for continuous database replication with minimal downtime; Glue cannot perform ongoing CDC from SQL Server to RDS. Option C is wrong because AWS DMS alone can handle the migration and replication, but without AWS SCT, schema conversion issues (e.g., unsupported data types, deprecated features) may cause failures or require manual intervention, especially if the source and target SQL Server versions differ. Option D is wrong because native SQL Server backup and restore to RDS is a one-time, offline process that does not support ongoing replication and would require significant downtime to restore a 1 TB database.

1486
MCQhard

Refer to the exhibit. A database specialist is troubleshooting an issue where an application cannot connect to an RDS for MySQL instance using IAM database authentication. The application uses the database user 'db_user1'. The IAM policy shown is attached to the IAM role used by the application. What is the most likely reason for the connection failure?

A.The action 'rds-db:connect' is not allowed for RDS MySQL.
B.The policy should have 'Deny' effect instead of 'Allow'.
C.The resource ARN in the policy uses an incorrect RDS resource ID.
D.The database user name in the ARN must be 'admin', not 'db_user1'.
AnswerC

The RDS resource ID must be exactly 14 alphanumeric characters. The example has 18.

Why this answer

IAM database authentication for RDS MySQL requires the resource ARN in the IAM policy to include the correct RDS resource ID (the 'db-xxxxx' identifier from the RDS console), not the DB instance name or endpoint. If the ARN uses an incorrect resource ID, the policy will not match the target RDS instance, causing the authentication to fail even if the user name and action are correct.

Exam trap

The trap here is that candidates often confuse the DB instance name or endpoint with the RDS resource ID, or assume the database user must be 'admin' for IAM authentication, when in fact the resource ID is a separate identifier and the user name must match the database user exactly.

How to eliminate wrong answers

Option A is wrong because the 'rds-db:connect' action is specifically allowed for RDS MySQL when using IAM database authentication; it is the required action for connecting. Option B is wrong because a 'Deny' effect would explicitly block the connection, whereas the goal is to allow it; the 'Allow' effect is correct for granting access. Option D is wrong because the database user name in the ARN must match the actual database user (here 'db_user1'), not 'admin'; the ARN format includes the database user name as it exists in the MySQL instance.

1487
MCQmedium

A development team is using Amazon RDS for MySQL with read replicas to offload reporting queries. They notice that the read replica is consistently lagging behind the primary by several seconds. The primary handles 5000 writes per second. Which action would most likely reduce replica lag?

A.Increase the 'max_connections' parameter on the primary instance.
B.Increase the instance size of the read replica.
C.Disable binary logging on the primary instance to reduce I/O.
D.Convert the primary instance to a Multi-AZ deployment.
AnswerB

A larger replica can apply changes more quickly, reducing lag.

Why this answer

Increasing the instance size of the read replica provides more CPU and memory resources, allowing it to apply changes from the binary log faster, thereby reducing replica lag. Option A is incorrect because increasing 'max_connections' on the primary does not affect the replica's ability to apply changes; it only allows more connections to the primary. Option C is incorrect because disabling binary logging on the primary would break replication entirely, as the replica relies on the binary log to receive changes.

Option D is incorrect because converting the primary to Multi-AZ improves availability and failover but does not directly reduce replication lag on the read replica.

1488
Multi-Selecteasy

Which TWO AWS services can be used to centrally manage database credentials securely? (Choose two.)

Select 2 answers
A.AWS Secrets Manager
B.AWS CloudFormation
C.Amazon S3
D.AWS Identity and Access Management (IAM)
E.AWS Systems Manager Parameter Store
AnswersA, E

Managed service for secrets.

Why this answer

Options A and E are correct. AWS Secrets Manager is specifically designed to centrally manage secrets, including database credentials, with features like automatic rotation. AWS Systems Manager Parameter Store can also securely store secrets (e.g., using SecureString parameters).

Option B (AWS CloudFormation) is incorrect because it is an infrastructure-as-code service, not a secrets manager. Option C (Amazon S3) is incorrect because it is an object storage service, not designed for secret storage. Option D (AWS IAM) is incorrect because it manages access permissions and identities, not secrets.

1489
MCQhard

A company is using Amazon DynamoDB with auto scaling enabled. The table's read capacity is set to a minimum of 100 and maximum of 1000 read capacity units (RCUs). The actual consumed read capacity is consistently at 200 RCUs. What should the database specialist do to optimize costs without impacting performance?

A.Lower the minimum read capacity to 200 RCUs.
B.Decrease the maximum read capacity to 500 RCUs.
C.Disable auto scaling and set the read capacity to 200 RCUs.
D.Increase the minimum read capacity to 500 RCUs.
AnswerA

Correct: Setting the minimum to 200 RCUs matches the actual consumption, avoiding paying for unused capacity while allowing auto scaling to handle spikes.

Why this answer

Lowering the minimum to 200 RCUs ensures auto scaling does not scale below the actual usage, reducing provisioned capacity costs. Option B is wrong because decreasing the maximum could cause throttling during spikes. Option C is wrong because disabling auto scaling would require manual management and may lead to over-provisioning or under-provisioning.

Option D is wrong because increasing the minimum would increase costs unnecessarily.

1490
MCQeasy

A company wants to migrate a 500 GB MySQL database from an on-premises server to Amazon RDS for MySQL. The migration must be completed within a 2-hour downtime window. Which migration approach is MOST suitable?

A.Use AWS Database Migration Service (DMS) with a full load.
B.Use mysqldump to export the database and then import it into RDS.
C.Create an RDS Read Replica of the on-premises database and promote it.
D.Create a compressed backup using Percona XtraBackup, upload it to S3, and restore it to RDS.
AnswerD

Percona XtraBackup creates a fast physical backup; restoring from S3 is efficient and can complete within 2 hours.

Why this answer

Percona XtraBackup allows you to create a physical backup of the MySQL database that can be compressed and uploaded to Amazon S3, then restored directly to an Amazon RDS for MySQL instance. This approach is significantly faster than logical backups (like mysqldump) for a 500 GB database, as it bypasses SQL parsing and rebuilds the data files directly, enabling completion within a 2-hour downtime window.

Exam trap

The trap here is that candidates often assume mysqldump (Option B) is the simplest and fastest method for a one-time migration, but they overlook the massive performance penalty of logical backups for large datasets, and they may not realize that Percona XtraBackup is the only option among these that can reliably complete a 500 GB migration within a 2-hour window.

How to eliminate wrong answers

Option A is wrong because AWS DMS with a full load alone does not meet the 2-hour downtime requirement for a 500 GB database; the initial full load can take hours depending on network bandwidth and database size, and DMS is designed for ongoing replication, not a one-time fast migration with strict downtime. Option B is wrong because mysqldump performs a logical export that serializes all data into SQL statements, which is extremely slow for 500 GB due to parsing, network transfer, and import overhead, often exceeding the 2-hour window. Option C is wrong because you cannot create an RDS Read Replica of an on-premises database; RDS Read Replicas only work within the RDS ecosystem (cross-Region or cross-AZ) and require an existing RDS source instance, not an external on-premises server.

1491
Multi-Selecthard

A company is designing a secure strategy for managing Amazon RDS for Oracle encryption keys. They want to use AWS KMS with Customer Master Keys (CMKs) for encryption at rest. Which THREE best practices should they follow?

Select 3 answers
A.Grant the RDS service principal (rds.amazonaws.com) only the necessary KMS permissions to use the CMK.
B.Create separate KMS keys for different environments (e.g., production, development).
C.Disable key rotation to maintain consistent encryption across all snapshots.
D.Store the KMS CMK inside the Oracle database for faster encryption.
E.Enable automatic rotation of the KMS CMK annually.
AnswersA, B, E

Least privilege ensures that only RDS can use the key for encryption operations.

Why this answer

Options A, B, and E are correct. Using separate KMS keys for different environments provides isolation (option B), enabling automatic key rotation is a security best practice (option E), and granting least privilege access to KMS keys is fundamental (option A). Option C is incorrect because disabling key rotation is not recommended and can lead to security risks.

Option D is incorrect because storing the CMK inside the database is insecure and defeats the purpose of using KMS.

1492
MCQmedium

A database specialist is troubleshooting an Amazon RDS for SQL Server instance that is running out of storage. The instance has 500 GB of provisioned storage and is using General Purpose SSD (gp2). The specialist wants to set up an alarm to notify when free storage space drops below 50 GB. Which CloudWatch metric and threshold should be used?

A.Monitor the 'FreeStorageSpace' metric in percent and set a threshold of 10.
B.Monitor the 'DiskSpaceUtilization' metric and set a threshold of 90.
C.Monitor the 'FreeStorageSpace' metric in bytes and set a threshold of 53687091200.
D.Monitor the 'FreeStorageSpace' metric in gigabytes and set a threshold of 50.
AnswerC

FreeStorageSpace is in bytes; 50 GB = 53687091200 bytes.

Why this answer

The correct metric is 'FreeStorageSpace' which is reported in bytes by CloudWatch for Amazon RDS. 50 GB equals 50 × 1024³ = 53,687,091,200 bytes. Therefore, setting a threshold of 53687091200 bytes triggers the alarm when free space drops below 50 GB. Option A is incorrect because 'FreeStorageSpace' is not available in percent; you must use bytes.

Option B is incorrect because 'DiskSpaceUtilization' is not a standard CloudWatch metric for RDS (it is used for EC2). Option D is incorrect because although 'FreeStorageSpace' is the right metric, the threshold must be specified in bytes, not gigabytes; CloudWatch does not accept a unit other than bytes for this metric.

1493
MCQhard

An IAM policy is attached to an application role that accesses a DynamoDB table named 'Orders'. The table has a global secondary index named 'OrderDateIndex'. The application needs to write new orders and query the index. Based on the exhibit, will the application be able to perform these operations?

A.Yes, but only writes are allowed; index queries are denied.
B.Yes, the policy allows both writes and querying the index.
C.No, the policy does not grant access to the index.
D.No, the policy denies Query on the index.
AnswerB

PutItem allowed on table, Query allowed on index.

Why this answer

The IAM policy grants `dynamodb:PutItem` on the table and `dynamodb:Query` on the index. Since the policy explicitly allows both actions on their respective ARNs, the application can write new orders to the 'Orders' table and query the 'OrderDateIndex' global secondary index. Option B is correct because the policy covers both required operations.

Exam trap

The trap here is that candidates assume a policy allowing actions on a table automatically extends to its global secondary indexes, but DynamoDB requires separate ARN entries for index-level operations like Query.

How to eliminate wrong answers

Option A is wrong because the policy explicitly allows `dynamodb:Query` on the index ARN, so index queries are not denied. Option C is wrong because the policy grants `dynamodb:Query` on the index ARN, providing explicit access to the index. Option D is wrong because the policy does not deny Query on the index; it allows it with an `Effect: Allow` statement.

1494
MCQmedium

A company is using an Amazon RDS for MySQL DB instance. The security team requires that all database connections be encrypted in transit. Which configuration step ensures this requirement is met?

A.Enable encryption at rest for the RDS instance.
B.Store the database password in AWS Secrets Manager.
C.Modify the network ACL to only allow traffic on port 3306.
D.Set the 'require_secure_transport' parameter to 1 in the DB parameter group.
AnswerD

This forces TLS connections.

Why this answer

Setting the 'require_secure_transport' parameter to 1 in the DB parameter group forces clients to use TLS/SSL for connections, ensuring encryption in transit. Option A is incorrect because enabling encryption at rest (RDS encryption) secures data on disk but does not enforce encryption of data in transit. Option B is incorrect because storing passwords in AWS Secrets Manager manages credentials but does not enforce encrypted connections.

Option C is incorrect because modifying a network ACL to allow traffic on port 3306 controls network access but does not enforce encryption of the data transmitted over that port.

1495
MCQeasy

A company is using Amazon DynamoDB with auto scaling enabled. The table's write capacity is set to a minimum of 50 and maximum of 500 write capacity units (WCUs). The actual consumed write capacity is consistently at 100 WCUs. What should the database specialist do to optimize costs without impacting performance?

A.Increase the minimum write capacity to 200 WCUs.
B.Disable auto scaling and set the write capacity to 100 WCUs.
C.Lower the minimum write capacity to 100 WCUs.
D.Decrease the maximum write capacity to 200 WCUs.
AnswerC

Matches actual consumption.

Why this answer

Lower the minimum write capacity to 100 WCUs. With auto scaling enabled and actual consumed write capacity consistently at 100 WCUs, the minimum should match the actual usage to avoid over-provisioning. Setting it to 100 WCUs ensures that baseline capacity is just enough for normal operations, while auto scaling can still scale up to 500 WCUs during traffic spikes.

Option A (increasing minimum to 200 WCUs) would raise costs without benefit. Option B (disabling auto scaling and fixing at 100 WCUs) removes the ability to scale up, risking throttling during spikes. Option D (decreasing maximum to 200 WCUs) limits scalability and could cause throttling if traffic exceeds 200 WCUs.

1496
MCQmedium

A company has an Amazon Redshift cluster that is running slowly on complex queries. The cluster has 10 dc2.large nodes. The 'QueryDuration' metric shows high values for several queries. The team wants to improve performance without changing queries. Which action is MOST likely to help?

A.Add more nodes of the same type to the cluster.
B.Enable compression on all columns.
C.Enable Redshift Spectrum to offload queries to Amazon S3.
D.Increase the workload management (WLM) concurrency level.
AnswerA

Adding more nodes increases the cluster's compute capacity, allowing complex queries to process more data in parallel without any changes to the queries themselves.

Why this answer

Adding more nodes (scaling out) increases the cluster's compute capacity, allowing complex queries to process data in parallel without any query changes. Option C is incorrect because Redshift Spectrum requires data to be in S3 and external tables to be defined, which would require modifying queries to reference those tables. Option B reduces storage I/O but not CPU-bound complex queries.

Option D increasing concurrency can cause resource contention, slowing queries.

1497
Multi-Selecthard

A company runs a production Amazon RDS for PostgreSQL instance with Multi-AZ deployment. The DB instance has a large number of connections from application servers. The operations team wants to monitor the number of database connections and receive an alert when it exceeds 80% of the maximum connections. Which combination of steps should be taken to set up this monitoring? (Choose two.)

Select 2 answers
A.Set the alarm threshold to 80 and the evaluation period to 5 minutes
B.Enable Enhanced Monitoring and create a CloudWatch alarm on the 'database_connections' metric
C.Set the alarm threshold to (0.8 * max_connections) and the evaluation period to 1 minute
D.Create a CloudWatch alarm on the 'DatabaseConnections' metric for the RDS instance
E.Enable Performance Insights and create a CloudWatch alarm using the 'DBLoad' metric
AnswersC, D

This ensures the alarm triggers when connections exceed 80% of the maximum.

Why this answer

Correct answers are C and D. C: Set the alarm threshold to (0.8 * max_connections) and the evaluation period to 1 minute ensures the alarm triggers when connections reach 80% of the maximum. D: Create a CloudWatch alarm on the 'DatabaseConnections' metric for the RDS instance to monitor the actual number of connections.

A is wrong because the threshold should be based on max_connections, not a fixed value of 80. B is wrong because Enhanced Monitoring provides OS-level metrics, not database connection counts. E is wrong because Performance Insights provides DBLoad, not connection counts.

1498
Multi-Selectmedium

A company uses Amazon DynamoDB with global tables. During a regional outage, the application fails over to the secondary region. After recovery, the DBA notices that the data in the secondary region is not fully consistent with the primary. Which THREE steps should the DBA take to diagnose the issue? (Choose THREE.)

Select 3 answers
A.Verify that DynamoDB Streams is enabled on the table.
B.Disable and re-enable global tables to force resync.
C.Increase the write capacity on the secondary table.
D.Review the DynamoDB Streams error logs in CloudWatch Logs.
E.Check the ReplicationLatency metric in CloudWatch.
AnswersA, D, E

Streams are required for global tables to replicate changes.

Why this answer

Options A, D, and E are correct. Verifying DynamoDB Streams (A) ensures that change data capture is enabled for replication. Reviewing DynamoDB Streams error logs in CloudWatch Logs (D) helps identify replication errors.

Checking the ReplicationLatency metric (E) reveals if there is replication lag. Option B is not a diagnostic step; disabling and re-enabling global tables is disruptive and not recommended. Option C is incorrect because increasing write capacity on the secondary table does not address consistency issues; replication depends on streams and network, not write capacity.

1499
MCQmedium

A company is designing a new application that requires a relational database with sub-millisecond read latency for a global user base. The workload is read-heavy with occasional writes. Which database solution should they choose?

A.Amazon DynamoDB with DAX
B.Amazon RDS for MySQL with Multi-AZ
C.Amazon Aurora with Auto Scaling
D.Amazon ElastiCache for Redis
AnswerC

Aurora provides low latency (single-digit ms) and is relational; Auto Scaling handles read scaling.

Why this answer

Amazon Aurora with Auto Scaling is the correct choice because it provides a relational database (MySQL/PostgreSQL-compatible) with sub-millisecond read latency via its distributed storage layer and read replicas. The read-heavy workload benefits from Aurora's automatic scaling of read capacity, while occasional writes are efficiently handled by the cluster volume. Aurora's architecture decouples compute and storage, enabling fast failover and consistent performance for global users.

Exam trap

The trap here is that candidates may confuse DynamoDB with DAX (which offers sub-millisecond latency) as a relational database, but DynamoDB is NoSQL and does not support relational features like joins or ACID transactions across multiple tables.

How to eliminate wrong answers

Option A is wrong because Amazon DynamoDB with DAX is a NoSQL key-value/document database, not a relational database, and while DAX provides microsecond latency for reads, the question explicitly requires a relational database. Option B is wrong because Amazon RDS for MySQL with Multi-AZ provides high availability but does not achieve sub-millisecond read latency; typical RDS read latency is in the single-digit milliseconds, and Multi-AZ is for failover, not read performance. Option D is wrong because Amazon ElastiCache for Redis is an in-memory cache, not a relational database; it can accelerate reads but does not serve as the primary relational database with ACID transactions and SQL querying.

1500
MCQeasy

A startup is building a real-time chat application that requires storing messages with high write throughput and low-latency reads. The data model is simple: each message has a conversation ID, timestamp, and content. Which database design is MOST appropriate?

A.Amazon RDS for MySQL with a single table and indexes on conversation_id and timestamp
B.Amazon Timestream to store messages as time-series data
C.Amazon Redshift with columnar storage and compression
D.Amazon DynamoDB with conversation_id as partition key and timestamp as sort key
AnswerD

This model supports high write throughput and efficient queries by conversation.

Why this answer

Amazon DynamoDB with conversation_id as partition key and timestamp as sort key is the most appropriate design because it directly supports high write throughput and low-latency reads for a real-time chat application. The partition key enables even distribution of writes across partitions, while the sort key allows efficient range queries for messages within a conversation ordered by time, matching the access pattern perfectly.

Exam trap

The trap here is that candidates often choose Amazon RDS for MySQL because they are familiar with relational databases and indexes, but they overlook the fundamental scalability limitations of a single-node RDS instance for high-write workloads, which DynamoDB's distributed architecture solves natively.

How to eliminate wrong answers

Option A is wrong because Amazon RDS for MySQL, while supporting indexes on conversation_id and timestamp, cannot scale to the high write throughput required by a real-time chat application without significant vertical scaling or complex sharding, and it introduces overhead from ACID transactions and locking that are unnecessary for this use case. Option B is wrong because Amazon Timestream is optimized for time-series data with regular intervals and aggregations, not for storing individual chat messages with high write throughput and low-latency point reads; it is designed for IoT and operational metrics, not real-time messaging. Option C is wrong because Amazon Redshift is a columnar data warehouse optimized for complex analytical queries on large datasets, not for high-frequency writes or low-latency point reads; its write performance is poor for transactional workloads, and it is not suitable for a real-time chat application.

Page 19

Page 20 of 23

Page 21