Questions d'entraînement

Question 1

The company stores many Amazon Machine Images (AMIs) that are critical. They need a solution to quickly recover accidentally deleted AMIs with minimal effort. Protect AMIs/snapshots against accidental deletion and enable easy recovery. Which solution meets the requirements?

Question 2

(Sélectionnez 2)

A healthcare organization operates 15 Linux-based research data servers across 3 different facilities. Each server contains critical patient research data with strict POSIX file permissions and symbolic links that must be preserved for compliance purposes. The organization needs to consolidate all research data into Amazon FSx for Lustre file system for high-performance computing workloads. POSIX permissions, symbolic links, and metadata must be preserved during migration. The total data size is approximately 500TB. Which solutions will meet these requirements? (Choose two.)

Question 3

A media company stores movies (1–10 GB files) in S3. Streaming must start within 5 minutes of purchase. Newer movies (<20 years) have higher demand than older ones. The company wants to minimize hosting cost based on demand. Select storage classes and retrieval that minimize cost while meeting a 5-minute availability target. Which solution meets the requirements?

Question 4

An e-commerce company has accumulated 15 TB of customer transaction data in Apache Parquet format stored in an Amazon S3 bucket over the past 3 years. The data includes purchase history, user behavior analytics, and seasonal shopping patterns. The marketing team needs to run SQL queries on this data monthly to generate business intelligence reports and analyze customer trends for strategic decision-making. Which solution will meet these requirements MOST cost-effectively?

Question 5

(Sélectionnez 2)

A healthcare research institute stores patient study data and clinical trial results in an Amazon S3 bucket. The data includes longitudinal research spanning 15 years and represents investments of over $50 million in research costs. The institute must ensure that research data cannot be accidentally deleted by researchers or administrative staff, as data loss would compromise ongoing studies and regulatory compliance requirements. Which combination of steps should a solutions architect implement to protect against accidental deletion? (Choose two.)

Envie de vous entraîner partout ?

Téléchargez Cloud Pass — inclut des tests d'entraînement, le suivi de progression et plus encore.

Question 6

A financial technology startup operates a real-time fraud detection system using Amazon EC2 instances. The system processes transaction data from an Amazon SQS queue, analyzing approximately 50,000 transactions during peak hours (9 AM - 6 PM) and 15,000 transactions during off-peak hours. The workload experiences sudden spikes during promotional events that can increase volume by 300-400%. The fraud detection system must maintain 24/7 availability with zero tolerance for downtime, as any interruption could allow fraudulent transactions to pass through undetected. The system needs to handle unpredictable traffic patterns while minimizing operational costs. Which solution meets these requirements MOST cost-effectively?

Analyse de la question

Core Concept: This question tests cost-optimized capacity planning for EC2 with variable and unpredictable workloads, balancing commitment-based discounts (Reserved Instances/Savings Plans) with burst capacity (On-Demand or Spot) while meeting strict availability requirements. Why the Answer is Correct: The system must be 24/7 with zero tolerance for downtime. That requirement strongly constrains the use of Spot Instances because Spot capacity can be interrupted with a 2-minute notice and can also be unavailable when you need it most (exactly during large promotional spikes). The most cost-effective approach that still preserves availability is to cover the steady, predictable baseline with Reserved Instances (or, in modern guidance, Compute Savings Plans) and then scale out for spikes using On-Demand Instances via an Auto Scaling group. On-Demand provides the highest reliability for burst capacity without interruption risk, ensuring the SQS backlog can be drained even during sudden 300–400% surges. Key AWS Features: Use an Auto Scaling group across multiple Availability Zones with scaling policies driven by Amazon SQS metrics (e.g., ApproximateNumberOfMessagesVisible, age of oldest message) to react to spikes. Purchase Reserved Instances (or Savings Plans) sized to the off-peak/steady-state processing level to reduce cost. Keep health checks, capacity rebalancing (if using mixed instances), and appropriate instance diversification to reduce capacity risk. This aligns with AWS Well-Architected Cost Optimization (use commitments for steady state) and Reliability (avoid interruption-prone capacity for critical workloads). Common Misconceptions: Spot looks “most cost-effective” for spikes, but it is not compatible with “zero tolerance for downtime” unless the workload is explicitly interruption-tolerant and designed for graceful degradation. Another misconception is reserving for maximum capacity; that wastes money because peak demand is limited to certain hours and promotional events are sporadic. Exam Tips: When you see requirements like “no downtime,” “mission critical,” or “zero tolerance,” avoid solutions that rely on Spot for required capacity. For variable workloads, the standard exam pattern is: commit to baseline with Reserved Instances/Savings Plans, burst with On-Demand for reliability (or Spot only if interruptions are acceptable). Also note SQS decoupling helps absorb spikes, but it does not remove the need for reliable compute to drain the queue within required time.

Question 7

A global e-learning platform company is implementing a multi-account strategy using AWS Organizations to provide isolated sandbox environments for their 50+ data science teams. Each team needs full administrative access to experiment with machine learning workloads and AWS services in their dedicated accounts. The security team requires that all API calls and activities must be logged through AWS Config for compliance auditing purposes. Since each data science team will have root-level permissions in their accounts, the architecture must prevent any team from disabling or modifying the mandatory AWS Config configuration that monitors critical security events. Which solution ensures that the mandatory AWS Config settings cannot be modified by the data science teams while maintaining their administrative access?

Analyse de la question

Core Concept: This question tests AWS Organizations governance controls—specifically Service Control Policies (SCPs)—to enforce mandatory security guardrails across member accounts, even when users have AdministratorAccess. It also touches AWS Config as a compliance/audit control and the need to prevent tampering. Why the Answer is Correct: An SCP applied to the organizational unit (OU) containing the sandbox accounts can explicitly deny actions that stop, delete, or alter AWS Config resources (e.g., StopConfigurationRecorder, DeleteConfigurationRecorder, PutConfigurationRecorder, PutDeliveryChannel, DeleteDeliveryChannel). SCPs define the maximum available permissions in member accounts; any explicit deny in an SCP cannot be overridden by IAM policies, including AdministratorAccess. Therefore, data science teams can retain full admin capabilities for experimentation while being technically prevented from disabling or modifying the mandatory AWS Config setup. Key AWS Features: - AWS Organizations SCPs: account-level guardrails that apply to all IAM principals in member accounts, including roles with admin permissions. - Explicit deny precedence: an SCP deny blocks the action regardless of IAM allow. - AWS Config organization-wide setup can be used, but the “cannot be modified by teams” requirement is met by SCP enforcement (often combined with org-wide Config for centralized management). - Best practice: combine preventive controls (SCP) with detective controls (AWS Config rules, CloudTrail) per the AWS Well-Architected Security Pillar. Common Misconceptions: - “Admin can do anything”: In Organizations, admin permissions are still bounded by SCPs. - “Root user can’t be restricted”: Root in member accounts is also subject to SCPs for API actions (though root can still perform some account-level tasks like closing the account). For the purpose of preventing Config API tampering, SCP is the correct control. - “Centralized Config alone prevents changes”: Organization-wide Config simplifies deployment, but without SCPs, admins in member accounts might still alter local Config components depending on how it’s deployed. Exam Tips: When you see “must prevent even admins from doing X across many accounts,” think SCP. IAM policies are not sufficient for cross-account governance, and service-linked roles don’t provide a governance boundary. For mandatory logging/configuration, pair org-wide services (Config/CloudTrail) with SCP guardrails that deny disabling or modification actions.

Question 8

A financial services company needs to deploy a real-time fraud detection system in the AWS Cloud. The system will analyze transaction data using machine learning algorithms running on 300+ Amazon EC2 instances that must simultaneously access shared training datasets totaling 50TB. The ML training workload requires sub-millisecond access latency to shared datasets for optimal model training performance. Multiple instances need concurrent read/write access to the same dataset files during distributed training. After training completion, data scientists need to access the processed model artifacts and training logs for analysis and validation. Which solution will meet these requirements?

Question 9

A global streaming media company operates over 250 video streaming platforms across different regions. The company needs to process approximately 25 TB of user viewing behavior data daily to optimize content recommendations and analyze viewing patterns. The solution must handle high-volume real-time data ingestion, provide reliable data transmission, and enable efficient large-scale analytics processing. What should a solutions architect recommend to ingest, transmit, and process the viewing behavior data?

Analyse de la question

Core Concept: This question tests designing a high-throughput, near-real-time streaming ingestion pipeline and landing data into a scalable analytics store. The key services are Amazon Kinesis Data Streams (durable stream ingestion), Amazon Kinesis Data Firehose (managed delivery to storage/analytics destinations), Amazon S3 (data lake), and Amazon Redshift (large-scale analytics). Why the Answer is Correct: With ~25 TB/day from 250+ platforms, the company needs elastic ingestion, reliable transmission, and efficient analytics. Kinesis Data Streams is purpose-built for high-volume real-time event ingestion with ordered, durable storage (retention) and horizontal scaling via shards. Kinesis Data Firehose then provides a fully managed, reliable delivery mechanism to continuously batch, buffer, optionally transform, and load data into an S3 data lake with minimal operational overhead. From S3, the data can be loaded into Amazon Redshift for performant SQL analytics at scale. This architecture cleanly separates ingestion, durable storage, and analytics, aligning with best practices for streaming data pipelines. Key AWS Features: - Kinesis Data Streams: shard-based scaling, multiple producers/consumers, retention for replay, enhanced fan-out for high read throughput. - Kinesis Data Firehose: automatic scaling, buffering (size/time), retries, compression, optional Lambda transformation, delivery to S3 (and other destinations). - S3 data lake: highly durable storage, lifecycle policies, partitioning for downstream query efficiency. - Redshift: columnar storage, MPP architecture, COPY from S3 for bulk loads; can be paired with Redshift Spectrum for querying S3 directly (often a follow-on design choice). Common Misconceptions: Options that “just store in S3 then process” miss the real-time ingestion and reliable transmission requirements. CloudFront is a content delivery/cache service, not an analytics ingestion buffer. DIY EC2 ingestion/processing can work but adds significant undifferentiated heavy lifting (scaling, fault tolerance, backpressure handling) compared to managed Kinesis services. Exam Tips: When you see “high-volume real-time ingestion” plus “reliable delivery” and “analytics,” think Kinesis (Streams for ingestion, Firehose for managed delivery) landing in S3, then query/warehouse with Redshift/Athena/EMR depending on the analytics need. Prefer managed services over custom EC2 fleets for streaming pipelines unless the question explicitly requires custom processing.

Question 10

A financial services company is operating a microservices-based trading platform using Docker containers orchestrated by Kubernetes in their private data center. The application uses a PostgreSQL database cluster for transaction data storage. Due to rapid business growth and increasing trading volumes, the company needs to migrate portions of their infrastructure to AWS. The migration must be completed within 6 weeks without any application code modifications or changes to the current CI/CD pipeline. The solution should minimize infrastructure management overhead while maintaining compatibility with existing Kubernetes manifests and PostgreSQL-based data layer. Which solution meets these requirements?

Practice Test #2

Réponses et explications vérifiées par triple IA

Questions d'entraînement

Témoignages de réussite(30)

Autres tests d'entraînement

Practice Test #1

Practice Test #3

Practice Test #4

Practice Test #5

Practice Test #6

Practice Test #7

Practice Test #8

Practice Test #9

Practice Test #10

Commencer à s'entraîner