Select Two (Multi-Select) Questions

Using two databases reduces costs by splitting storage between cheaper providers.

Why wrong: Cost reduction is not the rationale. Polyglot persistence chooses the optimal database type for each data access pattern — the appropriate tool for the job, which may actually cost more due to running two services.

Two databases provide automatic redundancy — if one fails, the other takes over.

Why wrong: Two different databases serving different workloads don't provide failover for each other. Redundancy is handled within each database service independently.

Use Cloud SQL's automated backups with a backup schedule every 4 hours.

Why wrong: Backup interval too long for RPO <5 minutes.

Store container images in a regional registry in the same region.

Why wrong: Does not help during a regional outage.

Use Cloud CDN to cache static content.

Why wrong: CDN is for improving latency, not disaster recovery.

Open source software

Why wrong: Open source does not inherently meet compliance regulations.

Public internet access

Why wrong: Public access increases risk, not a compliance feature.

Single data center footprint

Why wrong: Single data center is a single point of failure, not compliance.

High-frequency trading requiring microsecond latency

Why wrong: Cloud introduces network latency that may not meet HFT requirements.

Air-gapped, classified workloads with no internet connection

Why wrong: Cloud requires connectivity, so air-gapped workloads are not suitable.

Running a legacy mainframe application that requires dedicated hardware

Why wrong: Mainframes often need specific hardware not typically available in cloud.

Elimination of all security vulnerabilities

Why wrong: No system eliminates all vulnerabilities; cloud also has risks.

Complete transfer of security responsibility to the provider

Why wrong: Security is a shared responsibility; customer retains some controls.

Increased latency due to network distance

Why wrong: Latency is a performance trade-off, not a direct cost consideration.

Licensing costs for software that may not be compatible with cloud

Why wrong: This is a potential hidden cost but is not a standard cost-saving factor; it may increase costs.

Requirement to redesign applications for cloud-native services

Why wrong: This is an architectural consideration, not a direct cost comparison factor.

Replatform to App Engine Flexible Environment

Why wrong: App Engine Flex requires code modifications to fit the platform, which is not minimal change.

Refactor to microservices on GKE

Why wrong: Refactoring is a significant rewrite; not suitable for minimal changes initially.

Measuring everything with SLIs, SLOs, and SLAs

Why wrong: While measurement is important, the key principle is having service level objectives, not measuring everything.

Centralizing all operations in a single team

Why wrong: SRE advocates for shared ownership, not siloed operations.

Deploying changes only during maintenance windows

Why wrong: SRE encourages frequent, low-risk deployments, not restricted windows.

The maximum number of instances can be set to 'default' which is unlimited.

Why wrong: The default max instances is 100; you can set a higher value but there is no unlimited setting.

You can set a minimum number of instances to ensure zero cold starts.

Why wrong: Min instances reduce cold starts but don't guarantee zero; traffic bursts beyond warm instances can still cause cold starts.

Autoscaling uses CPU and memory utilization to make decisions.

Why wrong: Cloud Run autoscaling is based on request concurrency, not CPU or memory utilization.

Use a dedicated project for all PHI workloads

Why wrong: While recommended, it is not required; however, the question expects three correct, and E is not mandatory.

Configure multi-factor authentication for all users

Why wrong: MFA is a best practice but not specifically required by HIPAA.

Cloud Profiler

Why wrong: Cloud Profiler is for CPU and memory profiling of applications.

Cloud Debugger

Why wrong: Cloud Debugger allows inspecting code without stopping it, not for monitoring.

Cloud Trace

Why wrong: Cloud Trace is for distributed tracing to analyze latency, not general monitoring.

Virtualization

Why wrong: Virtualization is an enabling technology, not a NIST characteristic.

Multi-tenancy

Why wrong: Multi-tenancy is a common feature but not one of the five essential characteristics.

Serverless computing

Why wrong: Serverless is a model within cloud, not a defining characteristic.

Choosing the lowest-cost storage class

Why wrong: Cost does not relate to compliance.

Using a single availability zone for all data

Why wrong: Zone is redundant; data may still be in the wrong region.

Encrypting all data at rest

Why wrong: Encryption protects data but does not control location.

Deploy the database in a single region with backups

Why wrong: Backups alone don't ensure immediate failover; multi-region is required for regional failover.

Configure read replicas in a different region

Why wrong: Spanner does not use read replicas; it uses multi-region configs with automatic replication.

Use Cloud Memorystore to cache database queries

Why wrong: Caching is not a disaster recovery measure.

Cloud Operations Suite

Why wrong: Monitoring and logging, not part of CI/CD pipeline.

Cloud Deployment Manager

Why wrong: Used for infrastructure as code, not CI/CD pipeline.

Use spot VMs with a node taint and toleration for game server pods.

Why wrong: Spot VMs can be preempted at any time, causing game interruption, and are not suitable for stateful game servers.

Use preemptible VMs for game server pods.

Why wrong: Preemptible VMs can be terminated within 24 hours, causing game sessions to drop, which is unacceptable for real-time multiplayer games.

Disable the cool-down period to react faster to spikes.

Why wrong: Disabling cooldown can cause thrashing and unstable scaling.

Use a single metric (e.g., CPU utilization) for simplicity.

Why wrong: Using multiple metrics provides more stable and responsive scaling.

Scale based on custom metrics only.

Why wrong: Relying solely on custom metrics may miss system-level signals; using both is recommended.

Backend buckets for static content.

Why wrong: Backend buckets store static content but need CDN to achieve global low latency; alone they are regional.

Regional internal load balancer.

Why wrong: Regional load balancers are for traffic within a single region, not for global reach.

Query logs using Logging Query Language

Why wrong: Log queries are performed in Cloud Logging.

Automatically remediate incidents with Cloud Functions

Why wrong: Automatic remediation is not a built-in Monitoring capability.

IAP encrypts data at rest by default

Why wrong: IAP is for access control, not data encryption at rest.

IAP only works with Google Cloud applications

Why wrong: IAP can also protect on-premises applications via IAP for on-premises.

IAP requires using a third-party identity provider

Why wrong: IAP uses Google Workspace or Cloud Identity; it can also federate with third-party IdPs, but not required.