Skip to main content

Initial Setup

Updated Sep 21, 2020 ·

Overview​

Before creating any resources in Google Cloud Platform (GCP), spend 15–30 minutes securing your account and setting up cost controls.

This helps prevent:

  • Unexpected charges
  • Accidental resource creation
  • Cryptocurrency mining attacks
  • Forgotten resources running for weeks
  • Exceeding your free credits

Quick Checklist​

☐ Enable MFA

☐ Create a dedicated project

☐ Verify free trial credits

☐ Create a billing budget

☐ Configure billing alerts

☐ Review cost reports

☐ Learn the Always Free services

☐ Install the gcloud CLI

☐ Set your default project

☐ Use labels on resources

☐ Avoid creating unnecessary service account keys

☐ Delete unused resources after each lab

☐ Monitor running resources regularly

☐ Be cautious with high-cost services

☐ Review your billing dashboard weekly

1. Enable Multi-Factor Authentication (MFA)​

Your Google account is the root account for GCP.

Protect it first.

  1. Go to your Google Account. You can also go to:

    https://myaccount.google.com/security

  2. Open Security.

  3. Enable:

    • 2-Step Verification
    • Passkeys (optional)
    • Google Authenticator or hardware key

Never rely on password-only authentication.

2. Create a Dedicated GCP Project​

Avoid using the default project for everything.

Example:

my-lab

or

gcp-learning

Different projects make it easier to:

  • isolate resources
  • delete everything later
  • separate billing
  • manage IAM

3. Verify Billing Account​

Go to:

Billing

Verify:

  • Billing account is attached
  • Free trial credits are active
  • Credits are visible

4. Create a Billing Budget​

This is the most important step.

Go to Billing ➜ Budgets & alerts ➜ Create Budget

  1. Scope:

  2. Amount - You can set a smaller amount such as:

    $5

  3. Actions - Set the alert rules

    Recommended Alert Thresholds:

    • 25%
    • 50%
    • 75%
    • 90%
    • 100%

    Enable email notificationS, then click Finish.

    NOTE: This does not stop spending. It only sends alerts.

5. Budget vs Spending Limit​

Budgets do not stop charges.

Budget

  • Sends alerts
  • Does not stop resources

Quota

  • Limits service usage
  • Only applies to certain services

No global "maximum spend" exists in GCP.

Always monitor your usage.

6. Enable Cost Reports​

Go to Billing ➜ Reports

Learn how to view:

  • Daily spending
  • Service breakdown
  • Resource costs
  • Credits consumed

Check this page every few days.

7. Review Free Tier Services​

Google offers two types of free usage:

  1. Free Trial Credits (valid for a limited period)

    These credits can be used on many Google Cloud services.

    $300 credits

  2. Always Free Tier

    Certain services remain free within monthly limits.

    • Compute Engine (selected VM types in eligible regions)
    • Cloud Run
    • Cloud Functions
    • Cloud Storage
    • Firestore
    • Cloud Build

    Always verify the current Always Free limits before assuming a service is free.

8. Enable APIs Only When Needed​

Many services require enabling APIs.

Avoid enabling everything.

Enable only services you plan to use.

Examples:

  • Compute Engine API
  • Cloud Run API
  • Cloud Storage API
  • Cloud Functions API

9. Understand IAM​

Never work as the Owner account long-term.

Eventually create separate users or service accounts with least privilege.

For personal labs, Owner is acceptable initially.

10. Secure Service Account Keys​

Avoid downloading long-lived JSON keys unless absolutely necessary.

Prefer:

  • Workload Identity
  • Application Default Credentials
  • gcloud auth application-default login

If you create a key:

  • Store it securely
  • Rotate it regularly
  • Delete unused keys

11. Install Google Cloud CLI​

  1. Install gcloud CLI on your local machine.

    See Quickstart: Install the Google Cloud CLI

    Verify:

    gcloud version

    Sample output:

    joseeden@TOWER-1:Git$ gcloud version
    Google Cloud SDK 576.0.0
    alpha 2026.07.10
    beta 2026.07.10
    bq 2.1.34
    bundled-python3-unix 3.14.6
    core 2026.07.10
    gcloud-crc32c 1.0.0
    gsutil 5.37
    preview 2026.07.10

  2. Login:

    gcloud auth login

    It will return a URL. Open it in your browser and login.

    Back in the terminal, it should return a success message.

    You are now logged in as [*********************@gmail.com].
    Your current project is [None].  You can change this setting by running:
      $ gcloud config set project PROJECT_ID

  3. Set project:

    gcloud config set project PROJECT_ID

    To get the project ID, go to your GCP console, click on the project name, and copy the Project ID.

  4. Verify the current project:

    gcloud config list

    Sample output:

    [core]
    account = **************@gmail.com
    disable_usage_reporting = True
    project = project-********************

    Your active configuration is: [default]

12. Configure Regions Carefully​

Resources in different regions have different prices.

Choose a region close to you.

For Singapore:

asia-southeast1

Avoid creating duplicate resources across multiple regions unless needed.

13. Label Resources​

Apply labels from the beginning.

Example:

project=learning
owner=max
environment=lab

Labels help:

  • Identify resources
  • Filter costs
  • Simplify cleanup

14. Delete Idle Resources​

The biggest source of unexpected bills is forgetting resources.

Common examples:

  • Running VMs
  • Static external IPs
  • Persistent disks
  • Load Balancers
  • Cloud SQL
  • Kubernetes clusters

Always delete resources after finishing a lab.

15. Monitor Running Resources​

Useful pages:

  1. Compute Engine
  2. Cloud Storage
  3. Cloud Run
  4. Cloud SQL
  5. Kubernetes Engine

Regularly review what is currently running.

16. Enable Cost Recommendations​

Go to Cost Optimization

Google can recommend:

  • Deleting idle disks
  • Downsizing VMs
  • Removing unused IPs
  • Reducing costs

17. Set Quotas (When Available)​

Some services support quotas.

Examples:

  • CPUs
  • GPUs
  • API requests

Lower quotas reduce the risk of accidentally creating expensive resources.

18. Learn Which Services Become Expensive Quickly​

Be cautious with:

  • GPU instances
  • Large Compute Engine VMs
  • Kubernetes Engine (GKE)
  • Cloud SQL
  • Memorystore
  • BigQuery (large queries)
  • Load Balancers
  • Persistent SSD disks

These can consume credits quickly.

19. Create a Cleanup Habit​

After every lab session:

  1. Stop VMs.
  2. Delete unused disks.
  3. Delete unused snapshots.
  4. Delete test buckets.
  5. Delete test databases.
  6. Delete test load balancers.
  7. Review Billing reports.

These services are inexpensive or have generous free usage for learning:

  • Cloud Run
  • Cloud Storage
  • Cloud Functions
  • Secret Manager
  • Artifact Registry
  • Cloud Build
  • Cloud Logging
  • Cloud Monitoring
  • IAM
  • Compute Engine (small instances)
  • Cloud Shell

21. Services to Avoid Initially​

Until you're comfortable with pricing, avoid:

  • GPU instances
  • Large Compute Engine machines
  • Kubernetes Engine (GKE)
  • Cloud SQL
  • BigQuery with large datasets
  • Managed Redis (Memorystore)
  • Load Balancers
  • Multi-region storage
  • High-performance SSD storage