Refer to the exhibit. The application follows a containerized microservices architecture that has one container per microservice. The microservices communicate with each other by using REST APIs. The double-headed arrows in the diagram display chains of synchronous HTTP calls needed for a single user request.

Which action ensures the resilience of the application in the scope of a single user request?

Implement retries with exponential backoff during HTTP API calls.
Set up multiple instances of each microservice in active/active mode by using the Orchestrator.
Redesign the application to be separated into these three layers: Presentation, API, and Data.
Create two virtual machines that each host an instance of the application and set up a cluster.

Answer(s): A

Show Answer Next Question

QUESTION: 62

DRAG DROP (Drag and Drop is not supported)
Drag and drop the steps on the left into the order on the right for an end-user to access an OAuth2 protectedresource using the 'Authorization Code Grant' flow.

Select and Place:

See Explanation section for answer.

Answer(s): A

Explanation:

Show Answer Next Question

QUESTION: 63

In the three-legged OAuth2 authorization workflow, which entity grants access to a protected resource?

resource owner
client
resource server
authorization server

Answer(s): A

Reference:

https://developer.orange.com/tech_guide/3-legged-oauth/

Show Answer Next Question

QUESTION: 64

What are two steps in the OAuth2 protocol flow? (Choose two.)

The user is authenticated by the authorization server and granted an access token.
The user’s original credentials are validated by the resource server and authorization is granted.
The user indirectly requests authorization through the authorization server.
The user requests an access token by authentication and authorization grant presentation.
The user requests the protected resource from the resource server using the original credentials.

Answer(s): C,E

Reference:

https://www.digitalocean.com/community/tutorials/an-introduction-to-oauth-2

Show Answer Next Question

Viewing page 14 of 78
Viewing questions 66 - 70 out of 507 questions

Post your Comments and Discuss Cisco 350-901 exam dumps with other Community members:

Comments:

Name:

350-901 Exam Discussions & Posts

Anonymous User Commented on April 13, 2026
Question 12:

Correct answer: /sbin/init

Why: When using SysV init, the kernel starts the first user-space process with PID 1. That initial program is /sbin/init. It cannot exit and is responsible for starting the rest of the system (by reading /etc/inittab to determine the runlevel and starting the appropriate scripts in /etc/rc.d or /etc/init.d). The other options are not the initial program used by the kernel:

- /lib/init.so: a library, not an executable. - /etc/rc.d/rcinit: not the standard first program. - /proc/sys/kernel/init: not an executable program. - /boot/init: not a standard init program.

Quick note: On SysV systems, after /sbin/init starts, it spawns the runlevel scripts to bring the system to the desired state. Some systems may use /bin/init as a link to /sbin/init, but the key concept is that the kernel starts the initial user-space process.

Barueri, Brazil

Anonymous User Commented on April 13, 2026
Question 1:
For Question 1, the two correct options are A and D.

A. A rapid conversion of their existing SAP ERP/ECC environment to a modern, cloud-based architecture

- Why: RISE with SAP S/4HANA Cloud, private edition offers a faster, streamlined path to the cloud, reducing migration time and risk by moving to a modern cloud architecture.

D. Gaining the full Enterprise Management scope as a subscription

- Why: The offering provides the full scope of Enterprise Management available as a subscription, aligning licensing and ongoing updates with a cloud model.
Why the others aren’t correct here:

B (Reimagining processes and standardized best practices) is a benefit of transformation in general but not a specific migration advantage highlighted for this question.

C (Keep ECC 6.0 and minimize changes during migration) is not correct because you upgrade/migrate to S/4HANA; ECC 6.0 cannot be retained as-is in S/4HANA Cloud, private edition.

Gaborone, Botswana

Anonymous User Commented on April 13, 2026
Question 2:

This question tests how SmartConsole handles concurrent admin editing sessions on the Security Management Server.

When an admin opens SmartConsole, a separate editing session starts. Changes are local to that admin until they are published.

Other admins will see a lock on objects/rules that are being edited, but will not see the unpublished changes yet.

Therefore, if Jon is editing rule no.6 but has not published his changes, Dave will not see rule no.6 in his view. Dave only sees the rule once Jon publishes.

The correct answer is: D — Jon is currently editing rule no.6 but has not yet published his changes.

Glurns, Italy

Anonymous User Commented on April 13, 2026
Question 28:

Answer: B

Explanation:

- AWS Glue is a serverless data integration service ideal for creating data ingestion pipelines from data stored in S3. It handles ETL, clean/transform, and cataloging with minimal overhead. - Amazon SageMaker Studio Classic provides an integrated environment to build, train, and deploy ML models, making it suitable for creating the model deployment pipelines. - This pairing minimizes operational overhead: Glue handles ingestion and preparation of raw data, while Studio Classic handles deployment workflows and endpoints. - Why not other options: - Data Firehose is mainly for streaming ingestion, not batch ETL. - Redshift ML isn’t focused on ingestion pipelines. - Using a SageMaker notebook alone isn’t ideal for production deployment pipelines.
In short, use Glue for ingesting/transforming S3 data and Studio Classic for deploying the trained models.
Toyoda, Japan

ovelaguz Commented on April 13, 2026
Question 73:
The correct choices are C and D.

C: Replace the bastion host's SG to only allow inbound SSH (port 22) from the company’s external IP range. This ensures the bastion is reachable only from known on-premise networks, reducing exposure.

D: Replace the application instances’ SG to only allow inbound SSH from the bastion host’s private IP address. This enforces SSH access to apps only via the bastion over the private path inside the VPC.

Why these work:

They implement a secure bastion-based SSH flow: on-prem access goes to the bastion, then from the bastion’s private IP to the app instances.

Using the bastion’s private IP for the app SSH source prevents SSH from the bastion’s public endpoint, avoiding exposure to the internet.

Why the others are incorrect:

A: Limiting the bastion to inbound from application IPs blocks the initial SSH from on-prem to the bastion.

B: Allowing only internal IPs would block on-prem access via the internet to the bastion.

E: Allowing SSH to apps from the bastion’s public IP would expose the app tier to the internet and bypass the private path.

San Miguel De Allende, Mexico

ovelaguz Commented on April 13, 2026
Question 66:

Correct answer: C

Why: Use a lifecycle policy to move data from S3 Standard to S3 Standard-Infrequent Access (S3 Standard-IA) after 30 days. The data is frequently accessed in the first 30 days, then rarely accessed but still needs immediate retrieval. S3 Standard-IA provides cost savings after the initial period while preserving instant access, and it supports a 4-year retention.

Why the other options are not as good:

- A: Moving to S3 Glacier incurs retrieval latency; not ideal for immediate access needs in the first 30 days. - B: One Zone-IA has lower durability (Single AZ) and is riskier for critical data. - D: Moving to Glacier after 4 years adds unnecessary retrieval steps; Standard-IA already meets long-term, infrequent-access needs with immediate retrieval.
San Miguel De Allende, Mexico

ovelaguz Commented on April 13, 2026
Question 49:
Question 49: Explanation

Correct answer: B — Store individual files in Amazon S3 Intelligent-Tiering. Use S3 Lifecycle policies to move the files to S3 Glacier Flexible Retrieval after 1 year.

Why: Data under 1 year stays in a fast, automatically tiered storage class, while older data moves to a cheaper archival tier. This matches the pattern of frequent access for recent data and infrequent access for older data, with cost-effective analytics via Amazon Athena on current data and Glacier retrieval for older data when needed.

Why the other options are not suitable:

- A: Glacier Instant Retrieval isn’t ideal for ongoing analytics on data under 1 year; querying directly from Glacier is not efficient. - C: Relying solely on Athena without automatic tiering misses the cost optimization for data that ages into cheaper storage. - D: Overly complex (uses RDS for metadata) and uses Glacier Deep Archive, which is too slow for the near-term data and adds unnecessary components.

Key concept: Use a tiering strategy with S3 Intelligent-Tiering for active data and Glacier Flexible Retrieval for long-term archival, enabling cost-effective, fast access to recent data and inexpensive storage for older data.

San Miguel De Allende, Mexico

ovelaguz Commented on April 13, 2026
Question 48:
Question 48: Explanation

Correct answer: D — Move catalog to an Amazon EFS file system.

Why: EC2 instance store is ephemeral; data is not durable and is lost if the instance stops/terminates or fails. To achieve high availability and durability, you need shared, durable storage that multiple EC2 instances can mount.

Why other options are not suitable:

- A ElastiCache for Redis is in-memory caching, not durable storage for a catalog. - B Increasing instance size with instance store doesn’t provide durability or multi-AZ availability. - C S3 Glacier Deep Archive is archival storage with high latency, not suitable for a catalog requiring frequent access.

Key concept: Use Amazon EFS for a durable, scalable, shared filesystem that is accessible from multiple EC2 instances and replicated across AZs, delivering high availability and durability for the catalog data.

San Miguel De Allende, Mexico

Anonymous User Commented on April 13, 2026
Question 10:

Answer: AWS Lake Formation

Why this is correct:

- Lake Formation is a data lake governance service that helps centralize and catalog data from multiple sources (e.g., S3, on-prem databases via connectors), enabling unified datasets for analytics and ML. - It provides a single data catalog, centralized access control, and data preparation capabilities, making it easier to aggregate and prepare data from disparate sources for fraud-detection modeling.

Why the other options are not as suitable:

- Amazon EMR Spark jobs: Good for large-scale data processing, but not primarily for central data aggregation and governance across varied sources. - Amazon Kinesis Data Streams: Designed for real-time streaming data ingestion, not for batch data aggregation from multiple sources. - Amazon DynamoDB: A NoSQL database; not intended for integrating and aggregating heterogeneous training data from multiple sources.
Toyoda, Japan

Anonymous User Commented on April 13, 2026
SC-200 Learning Guide

Exam overview

- Domains and skills measured for Microsoft 365 Defender and security operations.

Module 1: Core Defender capabilities

- Defender for Endpoint: threat protection, remediation, device isolation, investigations. - Defender for Identity: monitoring and protecting domain controllers. - Defender for Cloud Apps (MCAS): anomaly detection, app control, blocked unsanctioned apps. - Defender for Office 365: phishing/aggregation protections, threat investigation.

Module 2: Azure Sentinel integration

- Create and configure analytics rules; automatic playbook (Logic Apps) execution.
Cape Town, South Africa

Aamir Commented on April 13, 2026
Great Content to understand the basics of terraform for exam point of view.
Islamabad, Pakistan

Anonymous User Commented on April 13, 2026
Question 128:
Answer: False
Explanation:

ACCOUNTADMIN is the highest-level admin role in Snowflake. Granting a newly created custom role to ACCOUNTADMIN would effectively extend that role’s privileges to the entire account, bypassing the principle of least privilege and jeopardizing separation of duties.

Best practice: create the custom role with only the specific privileges required, then grant that role to the appropriate users or to other roles that need those privileges. Reserve ACCOUNTADMIN for core admin tasks and avoid attaching additional privileges to it.

This approach improves governance and auditing, reduces the risk of privilege abuse, and makes privilege management more manageable.

Hyderabad, India

Anonymous User Commented on April 13, 2026
Question 119:

Answer: True.

Why:

- In Snowflake, a larger warehouse (e.g., 4X-Large) requires provisioning more compute resources (more nodes) than a smaller warehouse (X-Small). This extra provisioning work can take additional time. - If a warehouse is resized while running, Snowflake may need to reallocate resources, which can introduce a brief pause. - Provisioning time can also vary by cloud provider, region, and current load, so the larger size may occasionally take longer to provision than a smaller size.
Hyderabad, India

ovelaguz Commented on April 13, 2026
Question 193:

Correct answer: B. Use Amazon ElastiCache for Redis.

Why:

The goal is to reduce reads against RDS while keeping high availability. Using Redis as an in-memory cache sits in front of RDS, so frequently read data is served from memory instead of hitting the database.

ElastiCache for Redis supports replication and automatic failover, providing high availability for the cache layer itself.

Redis offers richer data structures and persistence options, making it more suitable than Memcached in this scenario.

Options A (EC2 MySQL) and C (Route 53 caching) do not reduce DB reads or provide appropriate caching/HA. D (Memcached) is an alternative but Redis is generally preferred for its features and resilience.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 189:
The correct answers are B and D.

B: Use S3 Object Lock in compliance mode. This enforces immutable, write-once behavior for the retention period, preventing any overwrites or deletions during the 5 years.

D: Use server-side encryption with AWS KMS customer managed keys (CMKs) and enable automatic key rotation. This provides at-rest encryption with automated rotation for long-term data.

Why the others aren’t correct:

A uses Object Lock in governance mode, which can be overridden; not guaranteed immutability for the full period.

C uses SSE-S3 (no key rotation control).

E uses imported keys, which require manual rotation and more overhead.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 188:
The correct answer is A.
Why A works:

They need a highly available, low-ops SFTP solution that writes to an S3 data lake. AWS Transfer Family provides a fully managed SFTP server with an S3 backend, handling scaling, availability, and server maintenance automatically.

It offers a public endpoint (or VPC endpoints if private access is needed) and integrates with IAM for authentication and access control, meeting partner onboarding and security needs without managing servers.

Why the others don’t:

B (S3 File Gateway) is for on-premises file access to S3 and does not natively provide an SFTP server for partners.

C–E require running and maintaining EC2 instances and custom upload logic (cron jobs), which adds operational overhead and is less available than a managed service.

Key concept: For a managed, highly available SFTP interface to S3 with minimal ops, use AWS Transfer Family with an S3 backend.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 172:

Answer: Configure a CloudFront field-level encryption profile.

Why this is correct:

- Field-level encryption (FLE) encrypts specific data fields at the CloudFront edge, so sensitive information stays encrypted throughout the request journey to the origin. - Only designated applications that hold the corresponding decryption keys can access the sensitive fields, meeting the requirement to restrict access.

Why the other options are not suitable:

- A) CloudFront signed URL: controls access to content, not per-field data encryption. - B) CloudFront signed cookies: also access control, not per-field encryption. - D) Origin Protocol Policy HTTPS Only: secures transport to the origin but does not encrypt specific fields or restrict field-level access.

Quick implementation idea (conceptual):

- Create a field-level encryption profile/config and specify which fields to encrypt. - Upload a public key to CloudFront for encryption; keep the private key with the designated applications (or your origin) for decryption. - Deploy so the origin can decrypt and handle data only by authorized apps.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 164:

Correct answer: C — Use Amazon SQS (standard queue) with a dead-letter queue, and integrate both applications with the queue.

Why this is correct:

SQS standard queue provides durable, highly scalable message storage and supports at-least-once delivery. Messages can be retained for up to 14 days, satisfying the requirement to keep messages if processing fails.

A dead-letter queue (DLQ) collects messages that fail repeatedly, preventing them from blocking others and enabling separate investigation.

Decoupling sender and processor improves operational efficiency; components scale independently and you avoid managing servers.

Why the other options are less suitable:

A: EC2 + Redis requires managing infrastructure and does not inherently guarantee durable persistence or robust message processing semantics.

B: Kinesis is a streaming service with more complex semantics and retention, and is not as straightforward for per-message retry with a DLQ.

D: SNS is publish/subscribe, not a durable queue; it can lose messages and lacks built-in dead-letter semantics for processing failures.

Key concepts:

Use a durable queue for decoupling and reliability.

Implement idempotent processing on the consumer side to handle potential duplicates from SQS.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 159:
Question 159 asks how to block unauthorized requests to a publicly accessible API built with API Gateway and Lambda when botnet traffic spikes.

Correct: A and C

- A: Create a usage plan with an API key shared only with genuine users. This restricts access to known users and allows quota controls at the API level. - C: Implement an AWS WAF rule to target malicious requests and filter them out at the edge before they reach Lambda.
Why the others aren’t correct

B: Filtering inside Lambda is brittle (IP spoofing, botnets) and adds processing cost; edge-based controls are more reliable.

D: Turning into a private API would block legitimate public users.

E: DNS redirection is not a proper access control measure and can disrupt users.

F: Creating an IAM role per user is not scalable for a public API.

Concepts:

Use API keys with a usage plan to control access to public APIs.

Use AWS WAF at API Gateway to block or rate-limit malicious traffic before it reaches backend services.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 157:
Question 157 asks how to meet 5-year data retention and indefinite audit logs for an Aurora PostgreSQL DB, given automated backups are already configured.

Correct: D and E

- D: Configure an Amazon CloudWatch Logs export for the DB cluster. This stores database activity/audit logs in CloudWatch Logs indefinitely, satisfying the requirement to keep audit logs long-term. - E: Use AWS Backup to take the backups and keep them for 5 years. AWS Backup provides centralized lifecycle management and can retain backups for a fixed 5-year period, meeting the data-retention requirement beyond what Aurora automated backups offer.
Why the others aren’t correct

A (Take a manual snapshot): not automated and not suitable for long-term retention.

B (Lifecycle policy for automated backups): not a defined feature for RDS/Aurora automated backups.

C (Automated backup retention for 5 years): automated backups in Aurora have a limited retention window (not indefinite 5 years) and don’t cover audit-log retention.

Concepts:

Use CloudWatch Logs exports to preserve audit logs beyond DB lifecycle.

Use AWS Backup for long-term retention and centralized policy control for backups.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 154:
The correct answer is B.

B. Use S3 Object Lock in compliance mode with a retention period of 365 days.

Why: Compliance mode enforces write-once-read-many for all objects for the retention period, and cannot be overridden, meeting the requirement that no one can modify or delete files for at least 1 year after creation. This also allows the few scientists to add new files, while existing files remain immutable for the retention window.
Why the others aren’t correct:

A. Governance mode with a legal hold of 1 year.

Governance mode can be overridden by users with special permissions; not truly immutable for all users as required.

C. IAM role to restrict deletions.

IAM alone cannot guarantee immutability or prevent overwrites/deletes once objects are added.

D. Lambda to track hashes.

Hash tracking does not enforce immutability or prevent deletions/modifications.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 150:
Answer: A — Create CloudWatch composite alarms where possible.

Composite alarms let you combine multiple underlying alarms (e.g., CPU > 50% and disk read IOPS high) and trigger only when all are in ALARM (AND logic). This matches the requirement to act only when both conditions occur, reducing false alarms from short CPU bursts.

Why not the others:

- Dashboards visualize data but don’t raise alarms or automate actions. - Synthetics canaries monitor availability, not real-time infrastructure metric correlation. - CloudWatch doesn’t support “single alarm with multiple thresholds” for correlated metrics; composite alarms are designed for this use case.
San Miguel De Allende, Mexico

Anonymous User Commented on April 12, 2026
Question 41:

Correct answer: Create an NS record named research in the adatum.com zone.

Why: Delegating a subdomain to another DNS server is done by adding an NS record in the parent zone (adatum.com) with the name of the subdomain (research) and pointing to the subdomain’s authoritative name servers. This tells resolvers to query the delegated server for research.adatum.com.

Why the others are wrong:

- PTR record is for reverse DNS, not delegation. - SOA record is the zone’s start of authority, not used for delegating a subdomain. - An A record for *.research would only map a wildcard host, not delegate authority.

Quick note: After creating the NS record, ensure the target DNS server is set up to serve research.adatum.com (likely with its own NS records) and register the appropriate NS at the registrar if needed.

Hyderabad, India

ovelaguz Commented on April 12, 2026
Question 143:

Answer: D: Host the application on Amazon ECS. Set up an Application Load Balancer with ECS as the target.

Why this is correct:

- You want to break the monolith into smaller, independently managed microservices with minimal operational overhead. ECS provides managed container orchestration, enabling multiple teams to own distinct services. - An Application Load Balancer routes traffic to the appropriate containerized services and supports auto scaling, enhancing scalability.

Why other options aren’t as suitable:

- A: Using Lambda is serverless but often requires substantial refactoring for a monolith moving to microservices; may introduce startup latency and stateful handling issues. - B: Amplify is frontend-focused and not designed to orchestrate multiple backend microservices. - C: Pure EC2 with ASG gives control but higher operational overhead vs a managed container approach for breaking into multiple services.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 142:

Correct answer: C

Why:

AWS Global Accelerator provides any-to-any UDP support, static IP addresses, and routes traffic to the nearest edge location, delivering the lowest latency for UDP-based traffic.

It can forward traffic to a Network Load Balancer (NLB), which is suitable for UDP/TCP workloads and supports static IPs.

Using EC2 instances in an Auto Scaling group gives control over kernel/UDP handling and allows scalable, highly available front-end servers.

Why the others aren’t as good:

A: Route 53 with ALB targets HTTP/HTTPS, not UDP, and lacks static edge IPs.

B: CloudFront is HTTP/HTTPS oriented and doesn’t support UDP at edge; NLB behind CloudFront adds complexity and latency.

D: API Gateway targets HTTP traffic; not suitable for UDP and doesn’t provide static edge IPs.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 141:

Correct answer: A

Why:

Use CloudFront with the ALB as the origin. CloudFront caches static content at edge locations worldwide, so users get static assets from nearby edges.

For dynamic content, CloudFront forwards requests to the origin (the ALB in the single region). The near edge location still reduces the round-trip time to the origin, improving latency for dynamic responses.

A single region is sufficient because edge caching at CloudFront delivers content globally without duplicating back-end deployments; latency is minimized by the edge network rather than multi-region routing.

Why the others aren’t as good:

B adds multi-region deployment and Route 53 latency routing, which is unnecessary since CloudFront already optimizes global delivery.

C only caches static content; dynamic content would still travel to the ALB and be slower.

D uses geolocation routing to a closest region but loses CloudFront’s global edge caching benefits and adds complexity.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 140:
Correct answers: A and C

A. Use Spot Instances for the data ingestion layer. The data ingestion on EC2 is sporadic and can tolerate interruptions, making Spot Instances the most cost-efficient option for this layer.

C. Purchase a 1-year Compute Savings Plan for the front end and API layer. The front end runs on Fargate and the API on Lambda. A 1-year Compute Savings Plan covers compute usage across EC2, Fargate, and Lambda, providing significant savings with flexibility across these services.

Why others are less optimal:

B On-Demand is more expensive for the ingestion layer than Spot.

D All Upfront RI for the ingestion layer is inflexible and doesn't suit potentially interrupted workloads.

E Savings Plan for EC2 only ignores Fargate/Lambda, which would miss savings on those services.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 139:

Answer: D

Why this is correct:

- Enable S3 Replication between the source and analysis buckets so new objects are automatically copied as they arrive, with minimal manual effort. - Use EventBridge to propagate ObjectCreated events from the analysis bucket to Lambda (for pattern matching) and to SageMaker Pipelines (for the ML pipeline), enabling event-driven processing.

Why the other options aren’t as good:

- A: Copies done by a Lambda function would add overhead, risk duplicates, and require custom logic instead of built-in replication. - B: Relying on EventBridge alone won’t ensure immediate cross-bucket replication. - C: Lacking EventBridge routing means coordinating Lambda and SageMaker without centralized event-driven triggers. - E: S3 replication is already the key; using a single root or assorting per-bucket rules adds unnecessary complexity.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 135:
Question 135 asks how to connect privately to a service hosted in an external provider’s VPC, with access restricted to that service and initiated only from your VPC.

Correct answer: D) Use AWS PrivateLink to connect to the target service. Create a VPC endpoint for the target service.

The provider creates a VPC Endpoint Service; you create an Interface Endpoint in your VPC to connect. Traffic stays on the AWS network and is restricted to that service.

Why:

- PrivateLink creates a private, service-specific connection that originates from your VPC and remains within the AWS network. - It limits access to the single target service, satisfying the security requirement.

Why the other options are incorrect:

- A: VPC peering connects entire VPCs, not a single service; traffic may reach other resources and isn’t PrivateLink-based. - B: A provider-side VPN/Gateway doesn’t restrict access to just one service and isn’t PrivateLink. - C: NAT gateway exposes traffic to the internet and does not establish a private, service-scoped connection.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 134:
Question 134 asks for a serverless, globally replicated data analytics solution for data in S3, with encryption and minimal ops.

Correct answer: C) Load into the existing S3 bucket. Enable CRR with SSE-S3. Use Athena to query the data.

Why:

- Using a single data store (S3) with serverless analytics (Athena) minimizes operational overhead. - Enable CRR to replicate encrypted objects to another region, fulfilling global availability and DR needs. - SSE-S3 keeps encryption managed by S3 with no extra key management.

Why the other options are less suitable:

- A: Requiring SSE-KMS multi-Region keys adds key management overhead and potential latency; although valid, it's unnecessary for least overhead. - B: Recommends RDS, which is not serverless analytics and increases operational overhead. - D: Also uses RDS, which introduces database management and is not aligned with a serverless analytics model.
Key concepts:

CRR replicates S3 objects between regions.

SSE-S3 provides server-side encryption with minimal management.

Athena enables serverless SQL queries directly on S3 data.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 133:
Question 133 asks how to upgrade an on-premises Oracle DB to the latest version, set up DR, minimize operational overhead, and keep OS access.

Correct answer: C) Migrate to RDS Custom for Oracle. Create a read replica for the database in another AWS Region.

Why:

- RDS Custom for Oracle provides managed DB provisioning with OS access for maintenance/admin tasks, while handling patching/ upgrades and reducing operational overhead. - A cross-region read replica gives DR capability with controlled lag and regional failover, meeting DR requirements while still allowing OS access for maintenance if needed.

Why the other options are less suitable:

- A) EC2 + manual replication: high operational overhead (full OS/db management, failover handling). - B) RDS for Oracle: no OS access; cross-region backups don’t provide OS-level control. - D) Standby in another AZ: not cross-region DR and still limits OS access.
Key concepts:

RDS Custom provides OS access plus managed DB operations.

Cross-region read replicas support DR with lower overhead than full active-active setups.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 131:
Question 131 asks how to serve all files via CloudFront while preventing direct access to the S3 URL.

Correct approach: Use an Origin Access Identity (OAI). Create an OAI, assign it to the CloudFront distribution, and configure the S3 bucket so that only the OAI has permission to read objects. This keeps the bucket private and ensures that objects can only be retrieved through CloudFront, not via direct S3 URLs.

Why the other options are incorrect:

- A: Creating per-object IAM users and policies for CloudFront isn't a standard, scalable way to restrict S3 access from CloudFront; CloudFront doesn’t use IAM users for access in this scenario. - B: A bucket policy with the CloudFront distribution ID as the Principal is not a valid pattern; OAIs are the supported mechanism. - C: A bucket policy that uses the distribution ID as Principal is invalid; you should use an OAI to grant S3 access to CloudFront.
Implementation outline:

In CloudFront, create an Origin Access Identity and attach it to the distribution.

In the S3 bucket, block public access and grant s3:GetObject to the OAI.

Verify that objects are accessible via the CloudFront URL but not via the S3 URL.

Outcome: S3 remains private; content is served securely and privately through CloudFront.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 120:
Question 120 summary:

Scenario: Self-managed DNS with three EC2s behind an NLB in us-west-2, plus another NLB in eu-west-1 with three EC2s. Goal: fast, highly available routing across US/Europe.

Correct answer: Create an Amazon Route 53 Global Accelerator with endpoint groups in us-west-2 and eu-west-1. Add the two NLBs as endpoints.

Why this is correct:

Global Accelerator provides a single fixed set of global IPs and health-aware routing across multiple regions, directing traffic to the closest healthy endpoint (here, the two NLBs). This improves performance and availability for users in both US and Europe with minimal configuration.

Why the other options are less suitable:

A: Geolocation routing + CloudFront is regional or cached and won’t automatically optimize health across regions, adding unnecessary complexity.

C: Attaching Elastic IPs to six instances is impractical and does not provide regional health-aware routing.

D: Latency-based routing to ALBs would require changing infrastructure and does not give centralized, global optimization like Global Accelerator.

Key concepts:

Global Accelerator uses endpoint groups per region and health checks to route traffic to healthy NLBs across regions.

It provides fast failover and consistent performance for multi-region setups.

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 117:

The requirement: store all application logs in Amazon OpenSearch Service in near real time with the least operational overhead.

Correct answer: A — Configure a CloudWatch Logs subscription to stream the logs to Amazon OpenSearch Service (OpenSearch). This is a fully managed, near-real-time integration with minimal setup and no custom code.

Why this is best:

- It provides near-real-time ingestion directly from CloudWatch Logs to OpenSearch. - It involves minimal configuration and no additional services or agents.

Why the other options are worse for this scenario:

- B: A Lambda function would require writing code, handling retries, scaling, and maintenance. - C: Kinesis Data Firehose adds an extra managed service layer and more configuration for near-real-time ingestion. - D: Installing agents on each server and using Kinesis adds significant operational burden and scaling concerns.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 109:
Answer: D
Explanation:

Use an S3 bucket with S3 Object Lock enabled and Versioning. This provides immutability for new uploads so objects can’t be deleted or overwritten during the retention/hold period.

Apply a Legal Hold to the objects. A legal hold prevents deletion or modification until it’s released, giving flexible control without a fixed retention period.

Grant delete capability only to specific users by adding the IAM permission s3:PutObjectLegalHold (and related policies) to those users, limiting who can remove the hold or delete objects.

Why the others don’t fit:

- A) Glacier is archival and not directly integrated with per-object delete permissions in S3. - B) Governance mode with a long retention prevents deletions but doesn’t provide per-user delete control. - C) CloudTrail/recovery does not enforce immutability.
San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 108:
Answer: A
Explanation:

This uses an event-driven pattern: when an RDS update occurs (a listing is sold), a Lambda function is triggered to publish a message to an SQS queue.

A standard (non-FIFO) SQS queue supports multiple consumers to poll and process the deletion data, enabling decoupled, scalable delivery to multiple targets with reliable delivery.

Why not B: A FIFO queue is unnecessary here; it introduces lower throughput and additional deduplication logic, which increases overhead.

Why not C: RDS event notifications are for DB instance-level events, not for per-row data changes like a sold listing; and adding SNS fan-out adds complexity.

Why not D: An SNS fan-out with multiple SQS queues adds extra hops; a direct Lambda-to-SQS flow is simpler and more efficient for this use case.

Key concepts:

Event-driven architecture with AWS Lambda and SQS

Decoupled, scalable delivery to multiple targets

Standard vs FIFO queues trade-offs (throughput vs strict ordering)

San Miguel De Allende, Mexico

ovelaguz Commented on April 12, 2026
Question 98:

Correct answer: C

Why this works:

- Standard SQS queues provide at-least-once delivery, so a message can be delivered again if processing isn’t complete before the visibility timeout. - Increasing the visibility timeout to exceed the sum of the Lambda function timeout and the batch window prevents the message from becoming visible again while still being processed, eliminating duplicate Lambda invocations and duplicate emails.

Why not the others:

- A) Long polling reduces empty receives but doesn’t prevent duplicates. - B) FIFO with deduplication is unnecessary overhead for this issue; Standard QoS already allows duplicates. - D) Deleting messages before processing risks losing messages if processing fails. - E) Not applicable to this scenario.

Key concept: Adjust the SQS visibility timeout in relation to the Lambda runtime and batch window to avoid reprocessing and duplicates.

San Miguel De Allende, Mexico

Anonymous User Commented on April 12, 2026
Question 86:
Question 86 asks for a secure way for web servers to connect to a common RDS MySQL Multi-AZ DB instance while meeting a requirement to rotate user credentials frequently.

Correct answer: A. Store the database user credentials in AWS Secrets Manager and grant the web servers the necessary IAM permissions to access Secrets Manager. Secrets Manager supports automatic rotation for RDS-compatible databases, giving centralized, secure, and frequently rotated credentials without manual effort.

Why this works:

Centralized, automatically rotated credentials reduce the risk of credential leakage.

Web servers fetch credentials securely from Secrets Manager via IAM permissions.

Rotation is built-in and scalable for multiple web servers.

Why the other options are not suitable:

B: AWS Systems Manager OpsCenter is for operational issue management, not credential storage or rotation.

C: Storing credentials in a secure S3 bucket requires manual rotation and lacks integrated rotation/auditability.

D: Per-host files encrypted with KMS do not provide centralized rotation or easy auditability for many servers.

Implementation hint:

Create a secret in Secrets Manager for the DB credentials, enable rotation for the secret, and attach an IAM role to the web servers to allow secretsmanager:GetSecretValue (and related permissions) to retrieve credentials at runtime.

San Miguel De Allende, Mexico

Anonymous User Commented on April 12, 2026
Question 13:
Yes. The correct answer for Q13 is C: Use seq on the loop on line 2.
Reason: for h in {1..254} relies on brace expansion and may not work in all shells. Using for h in $(seq 1 254); do improves portability and compatibility across POSIX-compliant shells, ensuring the script runs reliably in more environments.
Nixa, United States

Anonymous User Commented on April 12, 2026
Question 59:
Question 59 focuses on ingesting and analyzing 30+ TB of clickstream data daily.

Answer: D

Why: Use a managed streaming pipeline: collect with Kinesis Data Streams, deliver with Kinesis Data Firehose to an S3 data lake, then load into Amazon Redshift for analytics. This provides scalable, real-time ingestion and straightforward analytics loading.

Why the others are not suitable:

A) AWS Data Pipeline is deprecated for new workloads.

B) EC2-based ECS requires managing infrastructure and isn’t a managed streaming solution.

C) CloudFront is a CDN, not a data ingestion or streaming mechanism; Lambda alone isn’t scalable for continuous 30 TB/day without orchestration.

San Miguel De Allende, Mexico

Anonymous User Commented on April 12, 2026
Question 11:

Correct answer: A — Use AWS Secrets Manager with automatic rotation.

Why this is correct:

Centralizes database credentials in Secrets Manager instead of local files.

Supports automatic rotation for database credentials, reducing manual maintenance.

Applications can retrieve secrets at runtime without code changes, improving security and reducing ops overhead.

How to implement (high level):

Create a secret in Secrets Manager for the Aurora DB credentials.

Enable automatic rotation (MySQL/Aurora-compatible rotation) with the built-in Lambda function.

Grant your EC2 instances an IAM role that allows secrets:GetSecretValue.

Update the application to fetch credentials from Secrets Manager (or continue using the secret’s value without embedding passwords in the host).

Why the other options are less suitable:

B: Parameter Store can hold secrets but lacks the seamless, native rotation for database credentials you get with Secrets Manager.

C: Storing credentials in S3 and rotating with Lambda is manual and riskier; no native rotation flow.

D: Rotating via encrypted EBS volumes is complex, manual, and doesn’t provide centralized or automated credential rotation.

San Miguel De Allende, Mexico

Anonymous User Commented on April 12, 2026
Question 3:

Correct answer: A

Why A works:

- The global condition key aws:PrincipalOrgID allows you to grant access only to principals (users/roles) that belong to your AWS Organization. - By adding this condition to the S3 bucket policy, all accounts within the organization gain access without managing per-account permissions or OUs. - This minimizes ongoing admin effort and scales automatically as accounts are added to the org.

Why the others are less suitable:

- B (aws:PrincipalOrgPaths): Requires explicit OU paths and ongoing maintenance as accounts move between OUs; increases complexity. - C (CloudTrail monitoring): It logs events but cannot enforce real-time access control. - D (aws:PrincipalTag): Needs tagging every user and maintaining tag-based policies, adding manual overhead.

Quick example (conceptual):

- Policy with a Condition like "aws:PrincipalOrgID": "o-1234567890" grants access to any principal from your organization, with no per-user updates needed.
San Miguel De Allende, Mexico

Chad Commented on April 12, 2026
Question 6:

Correct answer: Emergency procedures.

Why: An MSDS (Material Safety Data Sheet) provides safety and handling information for hazardous materials. For a battery backup, it includes emergency procedures such as first-aid measures, fire-fighting steps, and spill containment.

Why the others are not correct:

- Installation instructions and Configuration steps are part of manuals or guides, not MSDS. - Voltage specifications are technical specs, not safety-focused data.
United States

Anonymous User Commented on April 12, 2026
Question 27:
Question 27 focuses on distributing connections to App1 across its VM hosts when users connect via different VPN methods.

Correct answers: A (internal load balancer) and E (Azure Application Gateway).

Why:

Internal load balancer: A private (internal) load balancer sits in the front-end subnet and spreads VPN-originated traffic across the VMs hosting App1, without exposing services publicly. This suits home-based P2S and site-to-site VPN traffic.

Azure Application Gateway: A layer-7 load balancer that can distribute HTTP/HTTPS requests across backend VMs, providing application-level routing and features like SSL termination.

Why the others aren’t suitable here:

Public load balancer would expose the app publicly, which isn’t required for VPN-based access.

CDN is for content delivery caching, not load balancing across VMs.

Traffic Manager is a global DNS-based load balancer across regions, not for balancing within a single region/VNet gateway setup.

In short, use an internal load balancer for private, VPN-based distribution, or an Application Gateway for app-level distribution.
Hyderabad, India

Anonymous User Commented on April 12, 2026
Question 153:

Correct answer: Perform a gap analysis to determine needed resources.

Why this is the FIRST action:

The organization already has non-compliance findings from internal audit. You need to understand exactly what is missing to meet regulatory requirements.

A gap analysis identifies the differences between the current controls/processes and the regulatory requirements, and it specifies the resources (people, processes, technologies, budget) needed to close those gaps.

Once gaps and resource needs are known, you can prioritize remediation and then perform a proper risk assessment to determine impact on business operations.

Other options are less appropriate as first steps:

- Create a security exception would bypass remediation and not address regulatory gaps. - Perform a vulnerability assessment targets weaknesses but not regulatory gaps or resource needs. - Assess the risk to business operations is important, but you need the gap/resource context first to accurately assess and prioritize risk.
Key concept: In governance, start with a gap analysis to map current state to regulatory requirements, enabling a actionable remediation plan and informed risk prioritization.
Ahmedabad, India

gw2fjrocha Commented on April 12, 2026
Question 11:

In Question 11, the statements are about three AI ethics principles: transparency, privacy, and inclusiveness.

- Box 1 (Transparency): Yes – transparency helps people understand how the model works. - Box 2 (Privacy): No – data must be protected; privacy is essential. - Box 3 (Inclusiveness): No – inclusiveness means AI should empower all people and remove barriers (e.g., accessibility features), not pricing or
Vagos, Portugal

Anonymous User Commented on April 12, 2026
what are the main topic in esam
Here are the main topics for the AI-900 exam:

Describe AI workloads and considerations — identify common AI workloads and factors like data privacy, security, ethics, fairness, and governance.

Describe fundamental principles of machine learning on Azure — basics of ML (supervised, unsupervised, reinforcement), model training and evaluation, and Azure ML services and workflows.

Describe fundamental concepts of computer vision workloads — tasks such as image classification, object detection, OCR, and related vision capabilities.

Describe fundamental concepts of NLP workloads and common use cases — topics like text analytics, translation, sentiment analysis, speech understanding.

Describe conversational AI workloads and Azure services — chatbots and virtual assistants, using services like Azure Bot Service, Language and Speech services, QnA.

Describe responsible AI — six principles: fairness, accountability, reliability and safety, privacy and security, inclusiveness, and transparency.

Hyderabad, India

Anonymous User Commented on April 12, 2026
Question 40:

The correct answer cannot be determined from the given information because the symbol rate S (in Gbaud/s) is missing.

Relationship: bits per symbol = 40 / S.

Possible cases:

- If S = 20 Gbaud/s ? 40/20 = 2 bits per symbol ? QPSK / 4QAM. - If S = 10 Gbaud/s ? 40/10 = 4 bits per symbol ? DP-QPSK. - If S = 40 Gbaud/s ? 40/40 = 1 bit per symbol ? BPSK.

Example: with S = 20 Gbaud/s, the modulation is QPSK (4QAM).

Cape Town, South Africa

jijore3700 Commented on April 12, 2026
Question 81:

Correct answer: Purchase Azure Active Directory Premium Plan 2 licenses for all users.

Explanation:

Azure AD Identity Protection features, including configuring a user risk policy and a sign-in risk policy, require Azure Active Directory Premium Plan 2.

Premium Plan 1 does not include Identity Protection, so upgrading to P2 is necessary to access these policies.

The other options don’t grant Identity Protection capabilities: MFA registration (B) isn’t the gating factor, security defaults (C) enable MFA but not risk policies, and Defender for Cloud features (D) are unrelated to Identity Protection.

Pfullingen, Germany

jijore3700 Commented on April 12, 2026
Question 80:

Role1 (manage application security groups): use the resource type Microsoft.Network/applicationSecurityGroups (i.e., the Microsoft.Network resource provider).

Role2 (manage Azure Bastion): use the resource type Microsoft.Network/bastions (i.e., the Microsoft.Network resource provider).

Explanation:

Both resources are network-related and live under the Microsoft.Network provider.

Role1 needs permissions on the Application Security Groups resource type; Role2 needs

Pfullingen, Germany

AI Tutor 👋 I’m here to help!

Cisco 350-901 Exam Questions Developing Applications using Cisco Core Platforms and APIs (DEVCOR) (Page 14 )

QUESTION: 60

QUESTION: 61

QUESTION: 62

Explanation:

QUESTION: 63

Reference:

QUESTION: 64

Reference:

350-901 Exam Discussions & Posts

Cisco 350-901 Exam Questions
Developing Applications using Cisco Core Platforms and APIs (DEVCOR) (Page 14 )