Microsoft AI-200 Exam Prep
Developing AI Cloud Solutions on Azure (Page 10 )

• The application failing to start or becoming unhealthy
• An incorrect probe path, port, or protocol
• The application needing more startup time
• Dependency or configuration failures
• A liveness probe being too aggressive

bash 
kubectl describe pod <pod-name> 
kubectl logs <pod-name> --previous

• Scale to more replicas: may hide the issue but does not diagnose it.
• Decrease initialDelaySeconds: makes probes run earlier and could worsen startup failures.
• Drain and reboot the node: node metrics are normal, so rebooting is not justified as the first step.

• Labels control access and protection, not blanket exclusion.
• If the user has permission to open a labeled file—especially if the label allows authenticated users to view it—Copilot can use that file in a summary.
• If the label applies encryption and the user lacks the required usage rights, Copilot should not use the file or reveal its contents.
• Copilot honors the user’s existing permissions, so it should not expose labeled content to an unauthorized user.

• The ERSPAN source is the device or interface whose traffic is being copied.
• The ERSPAN destination is the device receiving the encapsulated copy and forwarding it to the traffic analyzer.
• The local monitor session number is locally significant. Therefore, the source and destination do not need matching session IDs such as 47 or 48.
• The erspan-id identifies the ERSPAN stream and must correspond between the source and destination configuration.

• Crowd-sourced evaluation: Requires recruiting and managing many evaluators.
• Human-worker evaluation: Provides human judgment but requires coordination, labeling instructions, and quality control.
• RLHF: Is a training and alignment process, not simply a low-overhead model comparison method.
• Automatic model evaluation: Quickly evaluates outputs using predefined toxicity or quality metrics with minimal manual effort.

• Frequency/likelihood = how often or how likely the risk is to occur.
• Severity/impact = how damaging or costly the consequences would be.

• A minor equipment fault occurring frequently may have high frequency but low severity.
• A major cyberattack occurring rarely may have low frequency but high severity.
• A risk that happens often and causes substantial losses has high frequency and high severity.

• 255.255.255.128 = /25: 128 total, 126 usable
• 255.255.255.192 = /26: 64 total, 62 usable
• 255.255.255.224 = /27: 32 total, 30 usable
• 255.255.255.240 = /28: 16 total, 14 usable

• One for the network address
• One for the broadcast address

• 8 total addresses
• 6 usable addresses
• Mask: 255.255.255.248

• Honeypots: Decoy systems or services made to attract attackers. Legitimate users should not interact with them, so activity there is a strong indicator of malicious behavior.
• DNS sinkholes: Redirect requests for known malicious domains to a controlled internal system. This can reveal infected hosts attempting to contact command-and-control infrastructure.

• Tokenization: Replaces sensitive data with non-sensitive tokens.
• CI/CD: Automates software development, testing, and deployment.
• Threat modeling: Identifies and analyzes potential threats before or during system design.
• Data obfuscation: Makes data harder to understand or interpret.

• Microsoft Foundry is an enterprise platform for developers, data scientists, and IT teams.
• It supports the full AI application lifecycle: selecting models, connecting enterprise data, developing, evaluating, deploying, monitoring, and scaling solutions.
• It can include visual tools and low-code experiences, but it is not primarily designed to enable business users to build complete generative AI solutions independently.

• B: You still need to select and configure an appropriate model.
• C: Business users may participate through visual tools, but they are not the primary target audience.
• D: Foundry includes low-code tools, but it is not fundamentally a low-code-only platform.

1. x = -100
2. -100 < x < 100
3. 100 = x < 1000
4. x = 1000

• -500 ? class 1
• 0 ? class 2
• 100 ? class 3
• 1000 ? class 4

• Restrict a port to a defined number of MAC addresses
• Use statically assigned or “sticky” MAC addresses
• Shut down or restrict the port when violations occur

• Load balancer: Distributes application traffic; it does not protect switch MAC tables.
• IPS: May detect suspicious network activity, but is not the primary control for limiting switch MAC entries.
• NGFW: Filters traffic at the network boundary; it does not directly prevent MAC table flooding on access switches.

• Reviewing security objectives to ensure they support business goals.
• Ensuring security requirements are integrated consistently across business units.
• Coordinating security policies, risk management, controls, and awareness activities.

• Executive management: Provides direction, funding, and support, but usually does not manage day-to-day security integration.
• Board of directors: Provides governance and oversight, not operational security coordination.
• Steering committee: Helps coordinate and prioritize initiatives, but the CISO typically owns the overall security program and ensures its integration.

• Disaster recovery backups
• Long-term backup retention
• Data archiving that still needs occasional access

• Multi-Regional Storage: Intended for frequently accessed data needing broad geographic availability; it costs more.
• Regional Storage: Suitable for frequently accessed data within one region.
• Nearline Storage: Intended for data accessed roughly once per month, such as monthly backups.
• Coldline Storage: Better fit for backups accessed only during a disaster or occasional recovery event.

1. Discovery tries the SSH probe first.
2. SSH fails.
3. Discovery tries SNMP, which succeeds.
4. Discovery records SNMP as the preferred service for that IP.
5. Future discoveries try SNMP first.

• Classification identifies the type of device, such as a server, router, or printer.
• Credential affinity remembers which credentials work for a device.
• MID Server affinity associates a device or network with a particular MID Server.
• IP service affinity specifically controls the preferred discovery protocol or service.

• AWS Network Firewall can inspect and block traffic across a VPC when traffic is routed through the firewall. A Drop action rule can deny traffic from the suspicious source IP.
• GuardDuty findings are sent to Security Hub.
• A Lambda function can be invoked when Security Hub receives a new GuardDuty finding, extract the source IP, and update the Network Firewall policy with a blocking rule.

• GuardDuty threat lists: Used to identify or increase the severity of findings; they do not directly block VPC traffic.
• AWS WAF: Protects supported web applications, such as CloudFront or Application Load Balancers. It does not provide general VPC-wide traffic blocking.
• GuardDuty suppression rules: Suppress or hide findings. They do not block the suspicious traffic.

• Confirm the required columns and data types
• Identify the target/label column
• Handle missing or duplicate records if present
• Transform data into a suitable format
• Split it into training and validation/test datasets

1. Prepare the training data
2. Select an algorithm or configure AutoML
3. Train the model
4. Evaluate its performance
5. Tune hyperparameters if needed
6. Deploy the model

• Financial performance
• Customer perspective
• Internal business processes
• Learning and growth

• Increasing the number of indicators can make the scorecard cluttered and less useful. Measures should be limited to those that are strategically important.
• The scorecard is mainly a tool for implementing and monitoring strategy, not for formulating strategy.
• It should contain both leading indicators and lagging indicators. Leading indicators predict future performance, while lagging indicators measure results already achieved.

• A — Tables
• B — Named file formats
• C — Schemas
• E — Databases

Why these are correct

• Tables: Can be cloned using CREATE TABLE ... CLONE. The clone initially uses zero-copy cloning, so storage is shared until changes are made.
• Named file formats: Can be cloned as database objects, preserving the file-format definition.
• Schemas: A schema and its eligible objects can be cloned.
• Databases: A database and its eligible schemas and objects can be cloned.

Why the others are incorrect

• Shares: A share is an account-level data-sharing object, not a normal cloneable database object.
• Users: Users are security principals and cannot be cloned with CLONE.

• Label applied does not automatically exclude a file from Copilot.
• Encryption/usage rights determine whether Copilot can access the file’s content.
• If the user has no permission under Label2, File3 should be unavailable to Copilot for that user.
• File3’s label might be inherited by derived content only if File3 was actually processed and contributed by an authorized user; it should not be treated as contributing merely because it exists in the source set.

• Valor maior: mais aderência às instruções e aos detalhes descritos.
• Valor menor: mais liberdade criativa e resultados potencialmente mais variados.
• Valor excessivamente alto: pode causar artefatos ou imagens menos naturais.

• Aumentar o número de passos de geração: pode melhorar qualidade e refinamento, mas não é o principal controle de especificidade do prompt.
• MASK_IMAGE_BLACK: relaciona-se ao uso de máscara em geração ou edição de imagens, não à especificidade textual.
• Prompt strength: é mais associado à força do prompt em processos como image-to-image; nesta questão, o parâmetro diretamente ligado à aderência textual é o CFG scale.

1. Convert text and images into numerical embeddings.
2. Store or compare those embeddings.
3. Retrieve the most semantically similar items.

• Embedding model: Best for similarity search and retrieval.
• Generation model: Best for producing new text or other content from multimodal input.

• You can configure different pre-shared keys for different device groups.
• All cameras could use one key.
• All scanners could use another key.
• Each key can be associated with an appropriate role or access policy.

• Internal interface ? connected toward the corporate network/internal firewall
• External interface ? connected toward the Internet-facing firewall

1. A route for the internal corporate network through the internal-facing interface/firewall.
2. A default route (0.0.0.0/0) through the Internet-facing interface/firewall.

• Bronze: Raw ingested data
• Silver: Cleaned, validated, and transformed data
• Gold: Business-level data for analytics, dashboards, and reporting

• Owner, Contributor, and Network Contributor provide sufficient management permissions.
• Reader can view Traffic Analytics data but cannot enable or configure it.

• The identified risk
• Its likelihood and potential impact
• The person or team responsible for handling it
• The response or treatment plan
• Risk thresholds, such as when escalation or action is required
• Current status and review dates

• Risk tolerance: The amount of risk an organization is willing to accept; it is a concept or limit, not usually the primary documentation.
• Risk transfer: Moving the responsibility or financial impact of a risk to another party, such as through insurance.
• Risk analysis: The process of evaluating risk likelihood and impact.

• Storage usage of files stored in Microsoft Teams: SharePoint site usage report

• Number of active users per team: Microsoft Teams usage report

1. SharePoint site usage report
2. Microsoft Teams usage report

• Transfer — przeniesienie skutków lub kosztów ryzyka na inna organizacje, np. przez ubezpieczenie lub umowe.
• Akceptacja — firma swiadomie przyjmuje ryzyko bez dodatkowych dzialan.
• Mitigacja — zmniejszenie prawdopodobienstwa wystapienia ryzyka lub jego wplywu, np. przez wdrozenie kontroli bezpieczenstwa.
• Unikanie — calkowite wyeliminowanie ryzykownej dzialalnosci.

• A — Securing information assets: This is the responsibility of management and operational/security teams.
• B — Validating protection: Correct. The auditor checks whether controls—such as access restrictions, encryption, retention, and handling procedures—match the asset’s classification.
• C — Ensuring alignment with regulations: Management and legal/compliance functions typically establish classification requirements; the auditor may assess compliance but does not primarily define alignment.
• D — Defining classification levels: This is a management responsibility based on business needs, risk, and regulatory obligations.

• Limiting the number of MAC addresses allowed on a switch port
• Restricting ports to specific known MAC addresses
• Taking an action—such as dropping frames or shutting down the port—when the limit is exceeded

• Load balancer: Distributes application traffic; it does not protect switch MAC tables.
• IPS: May detect some attack traffic, but is not the primary control for switch-port MAC limits.
• NGFW: Filters network traffic at the firewall, but does not directly control MAC addresses learned by a switch.

• Microsoft Purview portal: Create and manage sensitivity labels and configure support for sensitivity labels used with Office files in SharePoint and OneDrive.
• SharePoint admin center: Primarily manages SharePoint settings. Older guidance and PowerShell procedures used SharePoint tenant configuration, which can make this question confusing.
• IRM: Concerns Information Rights Management and is a separate protection mechanism; it should not be used to identify the portal for configuring Microsoft Purview sensitivity labels.

• Accept: Acknowledging the risk without taking additional action.
• Transfer: Moving the risk’s financial impact to a third party, such as an insurer.
• Mitigate: Reducing the likelihood or impact through controls, such as patching or employee training.
• Avoid: Eliminating the activity that creates the risk.

• Deploy five individual VMs: This does not create a VM scale set.
• Modify Availability Zones or Size settings: These actions do not create the required scale set and add unnecessary steps.
• VM orchestration mode: This is Flexible orchestration, intended for more individually managed VMs and mixed configurations. It is not the best choice for a straightforward group of identical instances.

• A: Configure security groups — Customers define inbound and outbound rules for resources such as Amazon EC2 instances.
• B: Classify company assets — Customers decide how their data and assets should be classified and protected.

• C: Amazon S3 is a regional service; customers select a Region, not individual Availability Zones for buckets.
• D: AWS manages the infrastructure and patching for Amazon DynamoDB.
• E: AWS manages the infrastructure used by AWS Lambda; customers do not select EC2 instances for Lambda functions.

• Espionage: Stealing information for intelligence or competitive purposes.
• Data exfiltration: Unauthorized removal or transfer of data.
• Nation-state attack: An attack sponsored or conducted by a government.
• Shadow IT: Technology or access methods used outside approved IT processes.

• A. Periodically rotate responsibilities: May reduce some risks, but it does not directly detect unauthorized changes.
• B. Review segregation-of-duties conflicts: Identifies incompatible access combinations, not whether a specific data change was unauthorized.
• C. Log changes in an outside application: Logs provide evidence, but someone must review and analyze them. Logging alone does not detect misuse.
• D. Independent review: Directly detects unauthorized changes and reduces the risk that the person making the change can conceal it.

• Active remediation actions role in Microsoft Defender for Endpoint

• Security Reader role in Azure AD/Entra ID

• Security Administrator is broader than necessary because it grants extensive security-management permissions, violating least privilege.
• Compliance Data Administrator is intended for compliance data management, not Defender for Endpoint remediation approvals.

• Behavioral and anomaly detection
• Threat intelligence integration
• User and entity behavior analytics
• Automated response actions, often through SOAR integration

• Virus signatures: Associated primarily with antivirus or endpoint protection.
• Security agent deployment: Usually an endpoint-management or EDR function.
• Vulnerability scanning: Typically performed by vulnerability-management tools, though a SIEM may integrate their results.

• Fileless malware: Nothing in the command indicates malware executing only in memory.
• Living-off-the-land binaries: Mimikatz is a credential-access tool, not a normal system utility used for LOLBin abuse.
• Social engineering: The command shows technical credential abuse, not manipulation of an accountant.

spl 
error AND host=WWW3 
status=404 OR status=500 
NOT source=access.log

• Quotation marks are used for exact phrases or values containing spaces.
• Parentheses group parts of a Boolean expression, but they are not required for every Boolean search.

• The message came from the claimed sender (authentication)
• The message was not altered (integrity)
• The sender cannot easily deny sending it (non-repudiation)

• Attackers can send unsigned phishing emails.
• Attackers may compromise a legitimate account and send signed malicious emails.
• Users may still be tricked by a validly signed message.

1. Wireless AP

1. Router

1. Firewall

1. Device placement

• Each team’s Bedrock workflow uses a separate IAM service role.
• The role’s Amazon S3 permissions can be limited to only that team’s customer folder or objects.
• This enforces least privilege and prevents one team’s Bedrock application from accessing another team’s data.
• Access is enforced by AWS permissions, not by asking users to provide a customer name in each request.

• A single role with broad S3 access does not provide team-level isolation.
• Asking teams to specify a customer name is not an access-control mechanism.
• Redacting data does not prevent unauthorized access.
• A bucket policy alone is insufficient unless it explicitly enforces the required team-role restrictions.

• Employees can work from any location, including remote sites and offices worldwide.
• Login can be relatively seamless after the device is registered.
• It does not require distributing additional physical authentication equipment.

• Geotagging: Identifies location, but location alone is not a strong authentication method and may prevent legitimate remote access.
• Smart cards: Provide strong authentication but require card readers and physical cards, increasing equipment and management requirements.
• Time-based logins: Typically involve one-time codes that users must enter, making the experience less seamless and potentially requiring a token or authenticator app.

• Data Cloud combines data from multiple systems.
• It resolves and organizes that data into a unified customer view.
• That unified view enables downstream actions such as personalized campaigns, segmentation, analytics, and activation.

• A = core platform value
• D = a resulting business use case or benefit

• Confirming the data has the required columns and data types
• Handling missing or duplicate records
• Converting data into a suitable format
• Selecting or engineering useful features
• Splitting it into training, validation, and test sets

1. Prepare the training data
2. Select an algorithm or configure AutoML
3. Train the model
4. Evaluate its performance
5. Tune hyperparameters if needed
6. Deploy the model

1. Identify the fact table

• Records something that happened, such as a sale, test, appointment, or measurement
• Contains foreign keys to other tables
• Contains measures or numeric results
• Has many rows

2. Identify a dimension table

• Describe people, products, dates, locations, or departments
• Contain descriptive columns
• Are referenced by the fact table
• Have fewer rows than the fact table

3. Identify the surrogate key

• System-generated
• Unique within the dimension
• Often named PatientKey, CustomerKey, or similar
• Usually an integer
• Not meaningful outside the model

• Active-active routers: Provide redundant Internet connections/devices. Both can operate, so traffic can fail over or be load-balanced.
• Active-active firewalls: Avoid a single firewall becoming a failure point while maintaining security inspection.
• Load balancers: Distribute requests across multiple web servers. If one server fails, traffic can go to the other.
• Redundant power supplies: A server with separate power supplies should connect to independent power sources or circuits.
• UPS A and UPS B: Provide backup power and redundancy. Ideally, each power supply is connected to a different UPS.
• SONET ring: A ring topology provides alternate paths if a fiber link or segment fails.

• Domínios personalizados.
• Armazenamento suficiente.
• Instâncias dedicadas.
• Balanceamento de carga.
• Menor custo entre os planos que atendem a tudo.

• T?o m?t Systems Manager Automation runbook d? ch?y l?nh kh?i d?ng l?i ?ng d?ng trên các EC2 instance.
• T?o m?t Amazon EventBridge rule ph?n ?ng khi CloudWatch alarm chuy?n sang tr?ng thái ALARM.
• Ð?t Automation runbook làm target c?a EventBridge rule.

1. Log c?a ?ng d?ng du?c ghi vào CloudWatch Logs.
2. CloudWatch alarm phát hi?n nhi?u l?i liên quan d?n vi?c restart.
3. Alarm chuy?n sang tr?ng thái ALARM.
4. EventBridge nh?n s? ki?n thay d?i tr?ng thái này.
5. EventBridge g?i Systems Manager Automation runbook.
6. Runbook restart ?ng d?ng trên các EC2 instance, không restart c? instance.

• A: SNS không tr?c ti?p g?i Systems Manager Automation runbook nhu m?t target native.
• B: Lambda có th? th?c hi?n vi?c restart, nhung ph?i t? x? lý vi?c xác d?nh instance và ch?y l?nh t? xa.
• C: Lambda ch? làm trung gian d? g?i runbook, không c?n thi?t vì EventBridge có th? g?i runbook tr?c ti?p.