Google PCA Design and plan a cloud solution architecture Practice Question
An enterprise is migrating a latency-sensitive trading application from an on-premises data centre to Google Cloud. The application's components exchange hundreds of thousands of small messages per second and require sub-millisecond inter-process communication. The architect must choose a compute and networking design. What should the architect recommend?
⚠ Common exam trap
The trap here is optimizing for availability with multi-zone spread when the workload's dominant constraint is deterministic, sub-millisecond latency between components.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
Deploy the components on Compute Engine VMs in the same zone with compact placement, and enable high-priority network traffic using the `--network-performance-configs` total-egress-bandwidth-tier setting.
Ultra-low latency between tightly coupled components depends on physical proximity and adequate network throughput. Compact placement policies schedule instances close together on the same rack, reducing switch hops and jitter, while the higher total egress bandwidth tier removes the default throughput cap that would otherwise throttle a heavy small-message workload. Keeping all components in a single zone avoids inter-zone round trips entirely, which is essential for the stated sub-millisecond requirement.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
Deploy the components on GKE Autopilot pods spread across three zones with a PodDisruptionBudget and topology spread constraints.
Why it's wrong here
Spreading pods across three zones improves availability but forces inter-zone traffic for every message exchange, adding hundreds of microseconds to milliseconds of latency. Topology spread constraints and PodDisruptionBudgets address resilience, not speed. For a workload exchanging hundreds of thousands of small messages per second with a sub-millisecond budget, this design prioritizes the wrong attribute and cannot meet the latency target.
- ✗
Deploy the components on Cloud Run services in the same region and connect them through a Serverless VPC Access connector to a shared VPC.
Why it's wrong here
Cloud Run instances are managed and can be placed on shared infrastructure with cold starts and variable scheduling, and traffic through a Serverless VPC Access connector traverses an extra hop. That indirection is incompatible with a sub-millisecond messaging budget between tightly coupled components. Cloud Run is excellent for request-driven services, but it does not offer the deterministic, co-located networking this trading workload demands.
- ✓
Deploy the components on Compute Engine VMs in the same zone with compact placement, and enable high-priority network traffic using the `--network-performance-configs` total-egress-bandwidth-tier setting.
Why this is correct
Compact placement policies pack instances onto the same rack and physical network segment, which minimizes network hops between them and supports the low-latency, high-message-rate requirement. Setting the total egress bandwidth tier to the higher tier raises the VM network throughput ceiling so the small-message flood is not throttled. Keeping everything in one zone eliminates inter-zone round trips, making this the appropriate design for tightly coupled latency-sensitive components.
- ✗
Deploy the components on Compute Engine VMs in different zones of the same region and connect them with a global VPC using external IP addresses for direct communication.
Why it's wrong here
Placing components in different zones adds inter-zone latency that conflicts with sub-millisecond messaging, and routing traffic over external IP addresses exposes the application and adds path variability. A global VPC is the default scope for VPC networks, but global scope does not reduce latency between zones. This design increases both latency and attack surface, so it fails the performance requirement.
Go deeper
Related to this question
Learn chapter
Data Migration and Transfer Services
Key term
Data
Data is raw, unprocessed information, like numbers, words, or measurements, that can be stored, processed, and analyzed by computers.
Key term
Internet Protocol
Internet Protocol (IP) is the set of rules that governs how data is addressed, routed, and sent from one device to another across networks, including the internet.
About these practice questions
Courseiva writes every PCA question from scratch — 807 in total, each with an explanation and a wrong-answer breakdown. None are copied from real exams or dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official Google Cloud exam blueprint
This PCA practice question is part of Courseiva's free Google Cloud certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the PCA exam.