1 | Page
GOOGLE PROFESSIONAL CLOUD
ARCHITECT PRACTICE EXAM 2026/2027
QUESTIONS WITH VERIFIED ANSWERS &
COMPLETE RATIONALES
**1. Your company is migrating a legacy monolithic application
to Google Cloud. The application has unpredictable traffic
patterns with occasional spikes. The development team wants
to minimize operational overhead while maintaining the ability
to scale independently. Which compute service should you
recommend?**
A. Compute Engine with managed instance groups
B. Google Kubernetes Engine (GKE) Autopilot
C. Cloud Run
D. Compute Engine with custom machine types
**Correct answer:** C
**Rationale:** Cloud Run is a fully managed serverless platform
that automatically scales from zero based on traffic, requires no
cluster management, and charges only for resources
consumed during request processing. This minimizes
operational overhead while handling unpredictable traffic
spikes. GKE Autopilot (B) reduces management overhead
,2 | Page
compared to standard GKE but still requires cluster and
container orchestration knowledge. Compute Engine (A, D)
would require more operational management and wouldn't
auto-scale as seamlessly .
---
**2. Your organization is planning a multi-region deployment for
a global e-commerce platform. The application must maintain
99.99% availability and withstand the failure of an entire region.
Which architectural approach should you recommend?**
A. Single region with multiple availability zones
B. Active-passive with regional failover using Cloud DNS
C. Active-active with global load balancing and regional
redundancy
D. Single region with regional persistent disk replication
**Correct answer:** C
**Rationale:** Active-active with global load balancing provides
the highest availability by serving traffic from multiple regions
simultaneously. Cloud Load Balancing can distribute traffic
across regions with health checks, and if one region fails, traffic
automatically routes to healthy regions. Single region (A, D)
cannot withstand region failure. Active-passive (B) would have
a higher RTO .
,3 | Page
---
**3. Your company needs to migrate 50 TB of on-premises data
to Google Cloud Storage. The migration must complete within
30 days, and your network bandwidth is limited to 1 Gbps.
Which migration approach should you recommend?**
A. gsutil cp with parallel composite uploads
B. Storage Transfer Service
C. Transfer Appliance
D. Cloud Storage with multi-threaded uploads
**Correct answer:** C
**Rationale:** Transfer Appliance is a physical device that can
transfer up to 1 PB of data, bypassing network bandwidth
limitations. At 1 Gbps, transferring 50 TB would take
approximately 5–6 days at theoretical maximum, but in practice
with overhead, it could exceed the 30-day window. Transfer
Appliance provides a practical solution for large-scale
migrations where network bandwidth is constrained .
---
, 4 | Page
**4. A BigQuery data warehouse is experiencing slow query
performance. Queries scan entire tables even when filtering on
specific dates. What optimization should they implement?**
A. Add more slots
B. Implement table partitioning by date and clustering on
frequently filtered columns
C. Export to CSV and re-import
D. Use smaller tables
**Correct answer:** B
**Rationale:** Partitioning by date limits scans to relevant
partitions (date-filtered queries only scan those dates).
Clustering organizes data within partitions by specified
columns, improving filter performance. Combined, they
dramatically reduce data scanned and improve query speed.
More slots (A) add capacity but don't reduce data scanned .
---
**5. A company wants to establish SLIs, SLOs, and error
budgets for their application. They need to track availability and
latency, alert when SLOs are at risk, and make informed
decisions about feature velocity vs. reliability. Which approach
should they use?**
GOOGLE PROFESSIONAL CLOUD
ARCHITECT PRACTICE EXAM 2026/2027
QUESTIONS WITH VERIFIED ANSWERS &
COMPLETE RATIONALES
**1. Your company is migrating a legacy monolithic application
to Google Cloud. The application has unpredictable traffic
patterns with occasional spikes. The development team wants
to minimize operational overhead while maintaining the ability
to scale independently. Which compute service should you
recommend?**
A. Compute Engine with managed instance groups
B. Google Kubernetes Engine (GKE) Autopilot
C. Cloud Run
D. Compute Engine with custom machine types
**Correct answer:** C
**Rationale:** Cloud Run is a fully managed serverless platform
that automatically scales from zero based on traffic, requires no
cluster management, and charges only for resources
consumed during request processing. This minimizes
operational overhead while handling unpredictable traffic
spikes. GKE Autopilot (B) reduces management overhead
,2 | Page
compared to standard GKE but still requires cluster and
container orchestration knowledge. Compute Engine (A, D)
would require more operational management and wouldn't
auto-scale as seamlessly .
---
**2. Your organization is planning a multi-region deployment for
a global e-commerce platform. The application must maintain
99.99% availability and withstand the failure of an entire region.
Which architectural approach should you recommend?**
A. Single region with multiple availability zones
B. Active-passive with regional failover using Cloud DNS
C. Active-active with global load balancing and regional
redundancy
D. Single region with regional persistent disk replication
**Correct answer:** C
**Rationale:** Active-active with global load balancing provides
the highest availability by serving traffic from multiple regions
simultaneously. Cloud Load Balancing can distribute traffic
across regions with health checks, and if one region fails, traffic
automatically routes to healthy regions. Single region (A, D)
cannot withstand region failure. Active-passive (B) would have
a higher RTO .
,3 | Page
---
**3. Your company needs to migrate 50 TB of on-premises data
to Google Cloud Storage. The migration must complete within
30 days, and your network bandwidth is limited to 1 Gbps.
Which migration approach should you recommend?**
A. gsutil cp with parallel composite uploads
B. Storage Transfer Service
C. Transfer Appliance
D. Cloud Storage with multi-threaded uploads
**Correct answer:** C
**Rationale:** Transfer Appliance is a physical device that can
transfer up to 1 PB of data, bypassing network bandwidth
limitations. At 1 Gbps, transferring 50 TB would take
approximately 5–6 days at theoretical maximum, but in practice
with overhead, it could exceed the 30-day window. Transfer
Appliance provides a practical solution for large-scale
migrations where network bandwidth is constrained .
---
, 4 | Page
**4. A BigQuery data warehouse is experiencing slow query
performance. Queries scan entire tables even when filtering on
specific dates. What optimization should they implement?**
A. Add more slots
B. Implement table partitioning by date and clustering on
frequently filtered columns
C. Export to CSV and re-import
D. Use smaller tables
**Correct answer:** B
**Rationale:** Partitioning by date limits scans to relevant
partitions (date-filtered queries only scan those dates).
Clustering organizes data within partitions by specified
columns, improving filter performance. Combined, they
dramatically reduce data scanned and improve query speed.
More slots (A) add capacity but don't reduce data scanned .
---
**5. A company wants to establish SLIs, SLOs, and error
budgets for their application. They need to track availability and
latency, alert when SLOs are at risk, and make informed
decisions about feature velocity vs. reliability. Which approach
should they use?**
