GCP ACE · Domain 2 of 5

GCP ACE Domain 2: Planning and configuring a cloud solution

This section covers choosing the right Google Cloud services for a workload: selecting compute options, picking data storage and database products, and planning network resources such as load balancing and service tiers.

Exam format
~17.5%
Official sections
3
Assessment estimate
9–10
Suggested study time estimate
15–20 hrs

About 10 min read Reviewed October 2026

Overview

Planning and configuring a cloud solution covers selecting the right Google Cloud products for a workload before deploying them. It has three areas. Planning compute covers choosing among Compute Engine, Google Kubernetes Engine, Cloud Run and Cloud Functions, and using Spot VMs and custom machine types. Planning data storage covers choosing a database product such as Cloud SQL, BigQuery, Firestore, Spanner or Bigtable, and choosing storage options such as zonal or regional Persistent Disk and the Cloud Storage classes Standard, Nearline, Coldline and Archive. Planning networking covers load balancing, the availability of resource locations, and Network Service Tiers.

At about 17.5%, this section tests service-selection judgment: matching a workload's requirements to the right compute, storage, database and network choice. The exam rewards knowing each product's purpose and trade-offs rather than defaulting to a familiar option.

How it's tested

GCP ACE questions in this section describe a workload's compute, storage or network requirements and ask which product or option to choose. GCP ACE questions are practical and scenario-based, often involving a gcloud command, so read the requirement and choose the service or action it points to before answering.

  • Right compute choice

    A workload is described, and you choose Compute Engine, GKE, Cloud Run or Cloud Functions.

  • Right data product

    A data requirement is described, and you choose Cloud SQL, BigQuery, Firestore, Spanner, Bigtable or a storage class.

  • Network planning

    A connectivity or performance need is described, and you choose load balancing or a Network Service Tier.

Reasoning it through

A workload is a stateless container that must scale to zero when idle and run only in response to HTTP requests, without managing servers or clusters. Compute Engine requires managing VMs and GKE requires a cluster. Cloud Run runs stateless containers serverlessly, scaling to zero on no traffic and up on demand, which matches the requirement, so Cloud Run is the compute choice.

Sections 3

Official sections from the Google Cloud outline, in outline order.

2.1 Planning and configuring compute resourcesChoosing Compute Engine, GKE, Cloud Run or Cloud Functions, Spot VMs and custom machine types.
What it means
This covers selecting the appropriate compute choice for a workload among Compute Engine, Google Kubernetes Engine, Cloud Run and Cloud Functions, and using Spot VM instances and custom machine types where appropriate.
What to understand
Compute Engine gives full VM control, GKE runs containers on managed Kubernetes, Cloud Run runs stateless containers serverlessly with scale to zero, and Cloud Functions runs event-driven functions. Spot VMs offer deep discounts for interruptible work, and custom machine types tailor CPU and memory. The choice matches the workload.
How it shows up
A workload is described, and you must choose the compute product and options.
Common confusion
Confusing Cloud Run with Cloud Functions; Cloud Run runs containers, while Cloud Functions runs individual event-driven functions.
2.2 Planning and configuring data storage optionsDatabase product choice and Cloud Storage classes.
What it means
This covers choosing a database product such as Cloud SQL, BigQuery, Firestore, Spanner or Bigtable, and choosing storage options such as zonal and regional Persistent Disk and the Cloud Storage classes Standard, Nearline, Coldline and Archive.
What to understand
Cloud SQL is managed relational, BigQuery is analytics, Firestore is a document NoSQL store, Spanner is globally distributed relational, and Bigtable is wide-column NoSQL for large-scale low-latency workloads. Cloud Storage classes trade access cost against storage cost by access frequency. The choice matches the data model and access pattern.
How it shows up
A data requirement is described, and you must choose the database product or storage class.
Common confusion
Confusing Cloud SQL with Spanner; Cloud SQL is regional managed relational, while Spanner is globally distributed and horizontally scalable relational.
2.3 Planning and configuring network resourcesLoad balancing, resource location availability and Network Service Tiers.
What it means
This covers planning load balancing, the availability of resource locations such as zones and regions in a network, and choosing Network Service Tiers, the Premium and Standard tiers that trade performance against cost.
What to understand
Load balancing distributes traffic and comes in types by scope and protocol. Resource locations affect latency and availability, with zonal, regional and global choices. Network Service Tiers let you choose Premium, which routes over Google's backbone, or Standard, which is cheaper but uses the public internet more. The choice matches performance and cost needs.
How it shows up
A connectivity or performance need is described, and you must choose the load balancing or network tier.
Common confusion
Confusing the Premium and Standard Network Service Tiers; Premium uses Google's global backbone for performance, while Standard is lower cost.

High-yield concepts

The terms, rules and formulas this domain turns into questions most often.

  • Compute Engine

    Virtual machines with full control.

    Remember
    For lift-and-shift and specialized workloads.
    Tested as
    Compute questions.
    Trap
    Using it where serverless fits.

  • Google Kubernetes Engine

    Managed Kubernetes for containers.

    Remember
    For container orchestration at scale.
    Tested as
    Compute questions.
    Trap
    Using it for a single container.

  • Cloud Run

    Serverless stateless containers.

    Remember
    Scales to zero on no traffic.
    Tested as
    Compute questions.
    Trap
    Confusing it with Cloud Functions.

  • Cloud Functions

    Event-driven serverless functions.

    Remember
    For small event-triggered code.
    Tested as
    Compute questions.
    Trap
    Using it for container workloads.

  • Spot VMs

    Deeply discounted interruptible instances.

    Remember
    Only for fault-tolerant work.
    Tested as
    Cost questions.
    Trap
    Using them for critical workloads.

  • Cloud SQL vs. Spanner

    Regional managed relational versus globally distributed relational.

    Remember
    Spanner scales globally.
    Tested as
    Database questions.
    Trap
    Confusing the two.

  • BigQuery

    Serverless analytics data warehouse.

    Remember
    For large-scale analytics queries.
    Tested as
    Database questions.
    Trap
    Using it as a transactional store.

  • Firestore and Bigtable

    Document NoSQL and wide-column NoSQL.

    Remember
    Firestore for documents, Bigtable for scale.
    Tested as
    Database questions.
    Trap
    Confusing the two NoSQL stores.

  • Cloud Storage classes

    Standard, Nearline, Coldline and Archive by access frequency.

    Remember
    Match the class to access frequency.
    Tested as
    Storage questions.
    Trap
    Keeping cold data in Standard.

  • Persistent Disk options

    Zonal and regional block storage for VMs.

    Remember
    Regional adds zone redundancy.
    Tested as
    Storage questions.
    Trap
    Confusing zonal with regional.

  • Network Service Tiers

    Premium backbone versus Standard lower cost.

    Remember
    Choose by performance versus cost.
    Tested as
    Network questions.
    Trap
    Confusing the two tiers.

Side by side

Concepts candidates mix up, compared on the point that decides the answer.

Compute Engine vs. GKE vs. Cloud Run vs. Cloud Functions

ConceptMeaningKey differenceCommon trap
Compute Engine Full VM control Specialized, lift-and-shift Using it for serverless
GKE Managed Kubernetes Container orchestration Using it for one container
Cloud Run Serverless containers Scale to zero Confusing with Functions
Cloud Functions Event-driven functions Small event code Using it for containers

Cloud SQL vs. Spanner vs. BigQuery vs. Firestore vs. Bigtable

ConceptMeaningKey differenceCommon trap
Cloud SQL Regional managed relational Standard relational apps Using it for global scale
Spanner Global distributed relational Horizontal scale, strong consistency Confusing with Cloud SQL
BigQuery Serverless analytics warehouse Analytics at scale Using it transactionally
Firestore / Bigtable Document and wide-column NoSQL Documents vs. huge scale Confusing the two

Cloud Storage classes

ConceptMeaningKey differenceCommon trap
Standard Frequent access Highest storage cost Storing cold data
Nearline / Coldline Infrequent and rare access Lower storage, retrieval cost Using for frequent access
Archive Rarely accessed archives Lowest storage cost Using for active data

Exam traps

Each trap is an answer that looks right. Learn the clue that should change your mind.

  1. VM for serverless

    What candidates pick
    Use Compute Engine for a stateless workload that should scale to zero.
    Why it looks right
    VMs are a default choice.
    Why it's wrong
    Cloud Run runs stateless containers serverlessly and scales to zero on no traffic.

    Clue to watch for The workload is stateless and should scale to zero.

  2. Cloud SQL for global scale

    What candidates pick
    Use Cloud SQL for a globally distributed, horizontally scaled relational workload.
    Why it looks right
    Cloud SQL is managed relational.
    Why it's wrong
    Spanner is globally distributed and horizontally scalable relational; Cloud SQL is regional.

    Clue to watch for The workload is global and must scale horizontally with strong consistency.

  3. Standard for archives

    What candidates pick
    Store rarely accessed archival data in the Standard storage class.
    Why it looks right
    Standard is the default.
    Why it's wrong
    Coldline or Archive cost far less for infrequently accessed data.

    Clue to watch for The data is accessed rarely or is an archive.

  4. Spot for critical work

    What candidates pick
    Use Spot VMs for a workload that cannot tolerate interruption.
    Why it looks right
    Spot is cheapest.
    Why it's wrong
    Google can reclaim Spot VMs, so they suit only fault-tolerant, interruptible work.

    Clue to watch for The workload must not be interrupted.

  5. Functions for a container

    What candidates pick
    Use Cloud Functions to run a containerized application.
    Why it looks right
    Both are serverless.
    Why it's wrong
    Cloud Run runs containers; Cloud Functions runs individual event-driven functions.

    Clue to watch for The workload is a container rather than a single function.

Answer framework

For any Section 2 scenario, use these steps to find the answer the exam expects.

  1. Read the workload's requirements for compute model, data model, scale, access pattern and cost.
  2. For compute, choose Compute Engine for control, GKE for container orchestration, Cloud Run for serverless containers, or Cloud Functions for event-driven functions.
  3. For data, match the product to the model: Cloud SQL or Spanner for relational, BigQuery for analytics, Firestore or Bigtable for NoSQL, and the right Cloud Storage class for objects.
  4. For networking, choose the load balancing type and the Network Service Tier that match performance and cost needs.
  5. Use Spot VMs and custom machine types where appropriate, and confirm the choice fits the requirements before deploying.

Practice question

A team needs to run a stateless containerized web service that should automatically scale to zero when there is no traffic and scale up on demand, without managing virtual machines or a Kubernetes cluster. Which Google Cloud compute option should they choose?

Want more like this? The GCP ACE practice test has questions for every domain.

Checklist

0 of 10

Tick each one you can do without notes. Your ticks stay in this browser.

Study strategy

Approach this section as service selection, and build a decision map for each category. For compute, learn the four options and when each wins, especially that Cloud Run is for serverless stateless containers and Cloud Functions for event-driven functions. For data, learn the database products by model and scale, drilling the Cloud SQL versus Spanner and Firestore versus Bigtable distinctions, plus the Cloud Storage classes by access frequency.

For networking, learn the load balancing types at a high level and the Premium versus Standard Network Service Tiers. Because the exam gives requirements, practice reading a workload and recommending the product rather than memorizing feature lists.

Hands-on practice provisioning each service in a free-tier account cements the trade-offs. Because GCP ACE is hands-on and heavily tests the gcloud command line, practice every task in a free-tier account and the Cloud Shell rather than only reading, and review missed practice questions by section to find weak spots. Afterwards, move to Section 3, deploying and implementing a cloud solution, the largest section.

FAQ

How much of the GCP ACE exam is Section 2?

Planning and configuring a cloud solution is about 17.5% of the exam, focused on choosing the right compute, storage, database and network products.

What is the difference between Cloud Run and Cloud Functions?

Cloud Run runs stateless containers serverlessly, including scale to zero. Cloud Functions runs individual event-driven functions. Choose Cloud Run for containers and Functions for small event-triggered code.

What is the difference between Cloud SQL and Spanner?

Cloud SQL is a regional managed relational database for standard applications. Spanner is a globally distributed, horizontally scalable relational database with strong consistency, for large global workloads.

How do I choose a Cloud Storage class?

By access frequency: Standard for frequent access, Nearline and Coldline for infrequent and rare access, and Archive for rarely accessed data, trading storage cost against retrieval cost.

When should I use Spot VMs?

For fault-tolerant, interruptible workloads such as batch processing, where the deep discount outweighs the risk that Google may reclaim the instance.

What are Network Service Tiers?

Premium Tier routes traffic over Google's global backbone for better performance, while Standard Tier is lower cost and uses the public internet more. Choose by performance versus cost.

How does Section 2 connect to the rest of the exam?

The products chosen here are deployed in Section 3, operated in Section 4, and secured in Section 5, within the environment set up in Section 1.

Other Cloud Computing certifications

Where the same subject shows up in other exams. Cards marked "Same subject" open the matching domain.

Test your Domain 2 knowledge

You have covered choosing compute, storage, database and network products. Test whether you can match the product to a workload's requirements under exam conditions.

Take the GCP ACE practice test Check your GCP ACE readiness

Keep going