Amazon AWS Certified Solutions Architect - Associate SAA-C03 Practice Test 55

Question 540

Thousands of short-lived Lambda invocations each open a new TCP connection to an RDS MySQL database, exhausting backend connections. Which managed service can multiplex connections?

Increase the maximum allowable RDS database sessions but leave every Lambda invocation opening a new unpooled connection.
Add an RDS read replica and direct all write transactions to its read-only endpoint.
Enable Multi-AZ DB-instance failover and assume its synchronous standby multiplexes the primary’s client connections.
Place Amazon RDS Proxy between the Lambda clients and the database, then monitor connection pooling and avoid SQL session-pinning patterns that defeat reuse.

Explanation

Correct Answer

RDS Proxy can pool and multiplex backend MySQL database connections across many short-lived Lambda clients, reducing connection setup and memory pressure. Certain session state changes or long SQL statements can pin client sessions and reduce multiplexing, so monitor pinned sessions and keep appropriate authentication and pool settings.

Incorrect Answers

A larger limit may defer saturation but does not pool, reuse or multiplex the Lambda sessions that cause the connection explosion.

An RDS read replica does not accept normal write workload or pool thousands of short-lived MySQL client sessions.

The standby is for availability rather than client-side connection multiplexing; it does not solve primary-session churn.

Question 541

A PostgreSQL Aurora application has daily unpredictable peaks, a modest floor of traffic, and wants database capacity to scale without resizing provisioned instance classes. Which design fits?

Use a fixed, oversized provisioned Aurora DB class all month and assume the instance dynamically changes CPU allocation without scaling.
Use Aurora Serverless v2 with a tested minimum/maximum ACU range on a supported engine and Region.
Attach RDS Proxy to a fixed instance and assume the proxy adds database CPU capacity on every query spike.
Convert the relational schema to an S3 storage class and continue accepting the same PostgreSQL transactions there.

Explanation

Correct Answer

Aurora Serverless v2 scales writer/reader compute capacity within configured ACU boundaries as load changes.

Incorrect Answers

A provisioned instance has a defined compute class and does not provide the requested serverless ACU range-based elasticity by itself.

RDS Proxy pools connections and can improve resilience, but it does not make provisioned database compute capacity elastic.

S3 object storage does not implement the required relational PostgreSQL transaction processing semantics.

Question 542

An RDS workload uses gp3 storage and hits provisioned IOPS during a sustained write load although free disk space remains. Which action addresses the throughput limit?

Increase only RDS allocated storage from 200 GiB to 2 TiB without checking the provisioned gp3 IOPS and throughput settings.
Create a relational read replica and route the current write transaction stream to the reader endpoint.
Enable Multi-AZ deployment but leave the underlying provisioned gp3 IOPS bottleneck and write pattern unchanged.
Measure IOPS and throughput and increase the independently provisioned gp3 performance settings as appropriate.

Explanation

Correct Answer

gp3 volume performance can be provisioned independently of its storage GiB size; disk utilization does not capture IOPS saturation.

Incorrect Answers

gp3 decouples configurable IOPS and throughput from storage size in the supported ranges; free space alone is not the bottleneck.

Read replicas do not replace the writer for transactional updates, and the measured bottleneck is provisioned gp3 write I/O.

Availability replication does not by itself provision the additional storage I/O required for the sustained load.

Question 543

An RDS PostgreSQL query scans a large orders table for a selective customer filter and grew slow as records increased. Which diagnosis should precede buying a larger DB instance?

Double database CPU first and defer inspection of the table’s plan, predicate selectivity and index use.
Create a read replica while leaving the same SELECT plan unindexed on every copy.
Inspect EXPLAIN plans, filter selectivity and useful indexes before changing compute capacity.
Enable Performance Insights and assume the monitoring service automatically creates missing PostgreSQL indexes to correct every scan-heavy query.

Explanation

Correct Answer

A missing or unsuitable index may drive full scans, and query-plan evidence can distinguish a SQL access-path problem from CPU capacity.

Incorrect Answers

More CPU may mask symptoms but does not identify whether full scans, missing indexes or poor cardinality cause the growth-sensitive query.

Replicas can add read throughput but do not make an inefficient per-query table scan selective or efficient.

Performance Insights and execution-plan analysis help identify costly SQL and wait events; they do not automatically create an index or rewrite a selective query. Examine EXPLAIN and choose an access-path fix based on evidence.

Question 544

A DynamoDB inventory application must confirm that a successful write is visible to the next GetItem in the same Region before committing an order. Which read request behavior fits?

Read from a DynamoDB GSI with ConsistentRead=true and rely on the index to provide an immediately current value.
Use ConsistentRead=true on the base-table GetItem request in the same Region; use strongly consistent Query or Scan only where supported, such as on a local secondary index.
Issue an eventually consistent GetItem immediately after the write and assume it always sees the latest version.
Use DAX cached GetItem responses for transactions that require a guaranteed latest inventory count.

Explanation

Correct Answer

DynamoDB base-table GetItem can return the latest successfully committed value when ConsistentRead is true. Local secondary indexes also permit strongly consistent Query/Scan, but GetItem cannot target an index, and global secondary indexes offer only eventual consistency. The inventory lookup requires the base-table strongly consistent read.

Incorrect Answers

DynamoDB GSIs support eventual, not strong, consistency; specifying strong reads on a GSI is unsupported.

Eventual reads can temporarily return stale data even after a successful write acknowledgment.

DAX caches eventually consistent results; strongly consistent reads bypass cache, so cached answers cannot provide the requested guarantee.

Question 545

A worldwide application uses an Aurora Global Database with readers in Europe and primary writes in North America. Which performance benefit is the main reason for regional readers?

Assume every secondary Aurora Region can independently accept writes and reconcile them later using Multi-Region conflict merging.
Route all worldwide read requests to the primary writer Region to guarantee the minimum network round-trip latency.
Serve geographically local read-only traffic from secondary Regions while keeping writes coordinated at the primary.
Keep only one writer instance in North America and use a CloudFront object cache as a drop-in SQL transaction endpoint.

Explanation

Correct Answer

Global Database asynchronously replicates to secondary Regions, improving local read latency but not guaranteeing instantaneous freshness.

Incorrect Answers

Aurora Global Database generally has one primary write Region with asynchronous secondary read clusters; it is not a DynamoDB active-active model.

Remote users often incur higher latency when every read traverses the distance to the primary; regional readers reduce local read delay.

A CDN can cache HTTP responses, not replace the Aurora relational reader endpoints needed for regional read-only SQL.

Question 546

A business intelligence team performs large joins and aggregate scans on years of structured sales records. Which AWS managed database design is optimized for this analytics pattern?

Use a single provisioned RDS transactional instance and assume B-tree indexes alone make it an elastic petabyte analytical warehouse.
Load every year of sales records into DynamoDB and execute arbitrary SQL joins between tables using only partition-key operations.
Use Amazon Redshift as a columnar analytical warehouse with suitable data distribution and query design.
Use an OLTP-oriented Aurora Serverless v2 PostgreSQL instance and assume elastic CPU scaling alone creates a columnar massively parallel warehouse for repeated wide joins.

Explanation

Correct Answer

Columnar storage and massively parallel processing suit broad analytical scans and aggregates more than per-row OLTP access.

Incorrect Answers

An OLTP database is usually not tuned for large distributed columnar joins and warehouse-style aggregation at scale.

DynamoDB access patterns are key/value and indexed query oriented; it is not a direct substitute for large relational analytical joins.

Aurora Serverless v2 provides elastic relational compute, but its transactional execution engine is not a distributed columnar MPP warehouse. Redshift is purpose-built for large multi-table analytical aggregation workloads.

Question 547

A product catalog requires relevance-ranked free-text search over article titles and descriptions with aggregations by brand, not exact primary-key lookups. Which service is suited?

Index the documents in Amazon OpenSearch Service with appropriate analyzers and fields.
Use DynamoDB Query on the document ID as the only mechanism for relevance-ranked phrase search across free-text titles.
Put product descriptions into an S3 bucket and use bucket-name prefixes as a replacement for full-text indexing and aggregations.
Build an RDS transactional table and assume a primary-key lookup alone provides analyzer-based natural-language relevance ranking.

Explanation

Correct Answer

A full-text inverted index provides search relevance and aggregations beyond DynamoDB key-value query patterns.

Incorrect Answers

A partition-key Query is efficient for exact indexed access, but not a full-text relevance search engine with analyzers and scoring.

S3 object prefixes are storage namespaces rather than tokenization, relevance scoring or distributed text search.

Relational primary-key lookup does not supply the required analyzed free-text search and scoring semantics without a search-specific design.

Question 548

A DynamoDB checkout must decrement inventory and create an order atomically; accepting one write without the other causes overselling. Which feature meets this?

Write the order and decrement inventory in two ordinary PutItem calls and assume they both roll back if the second fails.
Use DynamoDB BatchWriteItem and assume every item operation is committed atomically with business-condition checks.
Use a DynamoDB GSI projection to join inventory and orders and automatically prevent overselling during concurrent updates.
Use DynamoDB transactional writes with conditions so all participating item changes commit or none do.

Explanation

Correct Answer

TransactWriteItems supports atomic coordinated changes across multiple items with conditional checks.

Incorrect Answers

Independent nontransactional writes can partially succeed; they do not provide all-or-nothing commit for the checkout invariants.

BatchWriteItem does not provide full multi-item ACID transaction semantics or conditional checks across the batch.

Indexes support query access, not atomic cross-item write coordination or business-rule condition enforcement.

Question 549

An Amazon Aurora cluster has three healthy Aurora Replica readers. A reporting service uses the Aurora reader endpoint but keeps every report on one long-lived pooled database connection; one reader remains hot and the others idle. Why does the reader endpoint not spread these individual queries evenly?

The reader endpoint always connects to the writer even when healthy Aurora Replicas exist in the DB cluster.
The reader endpoint rebalances each SQL statement across replicas over an existing database connection, so one hot replica means the replica role failed.
Aurora reader endpoints balance new client connections across readers; they do not migrate successive queries on a persistent connection between instances.
Add a fourth Aurora Replica but keep every report on the same existing pooled reader connection, expecting the new replica to accept part of each running SQL session.

Explanation

Correct Answer

A pooled long-lived session stays on its selected reader; distributing independent read connections, not merely many queries on one connection, is needed for load sharing.

Incorrect Answers

A reader endpoint with available Aurora Replicas routes compatible connections to those reader instances rather than always selecting the writer.

Connection balancing is not per-SQL-statement balancing; the SQL statements run on the server selected when that connection was established.

Adding replica capacity does not redistribute statements inside a persistent database connection. The reader endpoint balances new connections; separate or refreshed pooled read connections are needed to use additional readers.

Question 550

A reporting dashboard shows a newly committed order missing for several seconds on an RDS read replica, although the primary shows it. What explains the discrepancy?

The Multi-AZ standby temporarily stops synchronous replication whenever a SELECT query is issued against the source.
The source transaction was rolled back after its commit, and the replica always displays the last failed transaction.
A read replica is guaranteed to reflect each source commit before the source reports write success.
Asynchronous replication can introduce replica lag, so the replica need not reflect the latest primary commit immediately.

Explanation

Correct Answer

An RDS read replica is not a synchronous standby or a read-after-write consistency guarantee. Route immediate consistency reads to the primary.

Incorrect Answers

The scenario identifies a read replica; its asynchronous replication can lag even with no Multi-AZ standby changes.

A successful primary commit is not necessarily rolled back; an asynchronous replica can have outstanding changes yet to apply.

That would describe a synchronous consistency contract, which regular RDS read-replica asynchronous replication does not provide.

Pass Your Amazon AWS Certified Solutions Architect - Associate SAA-C03 Exam Easy!

Question 540

Explanation

Correct Answer

Incorrect Answers

Question 541

Explanation

Correct Answer

Incorrect Answers

Question 542

Explanation

Correct Answer

Incorrect Answers

Question 543

Explanation

Correct Answer

Incorrect Answers

Question 544

Explanation

Correct Answer

Incorrect Answers

Question 545

Explanation

Correct Answer

Incorrect Answers

Question 546

Explanation

Correct Answer

Incorrect Answers

Question 547

Explanation

Correct Answer

Incorrect Answers

Question 548

Explanation

Correct Answer

Incorrect Answers

Question 549

Explanation

Correct Answer

Incorrect Answers

Question 550

Explanation

Correct Answer

Incorrect Answers

AWS Certified Solutions Architect - Associate SAA-C03 Practice Tests