Oracle Cloud Free Tier for Learning

Khimananda Oli 7 min read Database
Oracle Cloud Free Tier for Learning

By Khimananda Oli | Last reviewed: August 2026

Most cloud providers offer free tiers that expire after 12 months or restrict you to micro-instances barely capable of running a static site. Oracle Cloud Free Tier for Learning breaks this pattern by providing four always-free ARM Ampere A1 compute instances with up to 24 GB of RAM and 200 GB of storage indefinitely. This makes it the most viable platform for building persistent DevOps labs, Kubernetes clusters, and database test environments without recurring costs. If you are preparing for certifications or prototyping infrastructure, understanding how to correctly provision and secure these resources is the first step toward effective cloud mastery.

What exactly does Oracle Cloud Free Tier for Learning include?

Unlike trial credits that vanish after 30 days, the Always Free resources in Oracle Cloud Infrastructure (OCI) remain active as long as your account is in good standing. The standout feature is the ARM-based compute capacity, which significantly outperforms comparable free offerings from AWS, Azure, or GCP. For engineers in Nepal or regions where high-performance cloud hardware is cost-prohibitive, this tier effectively provides a zero-cost enterprise-grade sandbox.

Oracle Cloud Free Tier ResourcesARM Ampere A14 OCPUs / 24 GB RAMFlexible Shape ConfigAlways FreeAMD E2.1.Micro2 Instances / 1 GB Eachx86_64 ArchitectureAlways FreeBlock Volume200 GB TotalUp to 5 VolumesBalanced PerformanceObject Storage10 GB StandardS3 Compatible APILoad BalancerFlexible LB10 Mbps BandwidthOutbound Data10 TB / MonthGlobal Egress Free
Oracle Cloud Free Tier for Learning resource allocation overview showing compute, storage, and networking limits

The ARM Ampere instances are particularly valuable because they support flexible shapes. You can allocate all 4 OCPUs and 24 GB RAM to a single instance for memory-intensive tasks like running PostgreSQL administration essentials, or split them into four smaller nodes for a Kubernetes cluster. The 200 GB block volume quota applies across all instances, so plan your partitioning carefully before provisioning. Note that while outbound data transfer is capped at 10 TB monthly, inbound traffic is completely unmetered—a critical advantage for backup-heavy workloads.

How do you provision and configure Always Free ARM instances?

A common mistake when using Oracle Cloud Free Tier for Learning is selecting the wrong shape during instance creation, which triggers paid billing. Always verify the "Always Free" label appears next to the shape selection. Here is the reliable workflow for provisioning an Ubuntu-based ARM lab server:

  1. Navigate to Compute > Instances and click "Create Instance."
  2. Select "Change Shape" and choose the VM.Standard.A1.Flex shape. Confirm the Always Free badge is visible.
  3. Configure OCPUs and Memory within the free limit (max 4 OCPUs, 24 GB RAM total across all A1 instances).
  4. Under Networking, ensure you select an existing VCN or create one with public subnet access. Save your SSH private key securely—OCI does not store it.
  5. In the Advanced section, add a cloud-init script to automate initial hardening. This aligns with practices from the initial Ubuntu server setup guide.
#cloud-config
package_update: true
package_upgrade: true
packages:
  - fail2ban
  - ufw
  - unattended-upgrades
runcmd:
  - ufw allow OpenSSH
  - ufw allow 80/tcp
  - ufw allow 443/tcp
  - ufw --force enable
  - systemctl enable fail2ban
  - systemctl start fail2ban

After provisioning, immediately configure security lists and network security groups (NSGs). OCI separates internet gateway routing from host-level firewall rules. Even if your NSG allows port 22, traffic will be blocked unless the subnet's route table includes an Internet Gateway and the security list permits ingress. Test connectivity with curl -v http://<public-ip> before assuming the instance is misconfigured.

How does Oracle Cloud Free Tier compare to AWS and Azure for labs?

When evaluating platforms for self-directed learning, raw specs matter less than sustained usability. Oracle Cloud Free Tier for Learning wins on persistence and compute density, but AWS and Azure offer broader managed service coverage. Understanding these trade-offs prevents frustration mid-project.

FeatureOracle Cloud Free TierAWS Free TierAzure Free Account
Compute DurationAlways Free (indefinite)12 months only12 months only
Max RAM (Free)24 GB (ARM)1 GB (t2.micro/t3.micro)1 GB (B1S)
Storage Included200 GB Block + 10 GB Object30 GB EBS + 5 GB S364 GB Managed Disk
Egress Allowance10 TB/month100 GB/month100 GB/month
Kubernetes SupportOKE (managed control plane free)EKS (paid control plane)AKS (free control plane)
Best ForPersistent labs, ARM-native appsShort-term trials, broad services.NET/Azure ecosystem learning

For Nepali developers and students, Oracle’s 10 TB egress is transformative. Transferring large datasets or serving media from a home lab won’t trigger surprise bills. However, if your goal is certification in a specific vendor ecosystem, use that vendor’s free tier despite lower specs. Real-world familiarity with IAM policies and CLI tooling outweighs raw horsepower when preparing for exams like the AWS Certified Solutions Architect.

How do you avoid unexpected charges and account suspension?

Oracle aggressively reclaims idle Always Free resources. Instances with less than 10% CPU utilization over seven consecutive days may be stopped automatically. To prevent this in learning environments, deploy lightweight background tasks that simulate baseline activity. A simple cron job writing timestamps to a log file every five minutes suffices. More importantly, set up budget alerts even for free-tier accounts. Navigate to Billing > Budgets and create an alert rule with a threshold of $0.01. This catches accidental paid-resource provisioning before costs accumulate.

Maintaining Always Free EligibilityDeploy InstanceVerify Always Free TagSet Budget Alert$0.01 ThresholdSimulate Activity>10% CPU Weekly AvgMonitorUsage DashboardCritical Guardrails• Never exceed 4 OCPUs / 24 GB RAM across ALL A1 instances• Avoid attaching paid boot volumes or upgrading shapes accidentally• Reclaim stopped instances within 7 days or lose attached IPs
Workflow for maintaining Oracle Cloud Free Tier for Learning eligibility and preventing resource reclamation

Account suspension often occurs due to verification failures. OCI requires valid payment method verification even for free tiers. Use a credit card (not prepaid/debit) and ensure your billing address matches bank records exactly. In Nepal, international cards from Nabil or Himalayan Bank typically pass verification, while some virtual cards fail. If suspended, open a support ticket immediately—automated reinstatement rarely works. Document your use case clearly: "Educational lab for DevOps certification preparation" resolves faster than vague descriptions.

What practical projects maximize learning value on this tier?

Raw infrastructure is meaningless without purposeful projects. Structure your Oracle Cloud Free Tier for Learning environment around real-world scenarios that build portfolio-worthy skills. Start with a three-node k3s cluster using the ARM instances. k3s runs efficiently on ARM and consumes minimal overhead, leaving room for workloads. Pair this with Longhorn for distributed storage to practice stateful application management, as detailed in the Longhorn distributed storage guide.

Next, implement observability. Deploy Prometheus and Grafana using Helm charts to monitor cluster health. The 24 GB RAM pool comfortably supports both tools plus sample applications. Create dashboards tracking the four golden signals—latency, traffic, errors, saturation—to internalize SRE principles. This hands-on experience directly complements theoretical knowledge from the four golden signals article. Finally, expose services via OCI’s free flexible load balancer with Let’s Encrypt TLS termination. Practice certificate automation with cert-manager to understand PKI workflows without risking production systems.

DevOps Learning Stack on OCI Free TierOCI Flexible Load Balancer (Free)TLS Termination • Path Routing • Health Checksk3s Cluster (3 ARM Nodes)Control PlaneNode 1: 2 OCPU/8GBWorker + MonitoringNode 2: 1 OCPU/8GBWorker + StorageNode 3: 1 OCPU/8GBLonghorn StorageDistributed PVsPrometheus + GrafanaMetrics • Dashboards • AlertsSample Apps + CI RunnersPortfolio Projects • GitLab Runner
Complete DevOps learning architecture using Oracle Cloud Free Tier for Learning with Kubernetes, monitoring, and persistent storage

Start Building Your Persistent Cloud Lab Today

Oracle Cloud Free Tier for Learning remains unmatched for persistent, high-capacity cloud experimentation in 2026. By provisioning ARM instances correctly, enforcing strict budget guardrails, and structuring projects around real operational challenges, you transform free credits into genuine engineering competency. Don’t let the resources sit idle—deploy a cluster today, break it intentionally, and document your recovery process. That documented journey is worth more than any certification badge. Ready to design your lab architecture or troubleshoot a stalled deployment? Reach out through my contact page for tailored guidance on maximizing your OCI learning environment.

Frequently Asked Questions

You get two AMD micro instances, four ARM Ampere A1 cores with 24GB RAM, and 200GB block storage. These Always Free resources do not expire and remain available indefinitely for learning DevOps, Linux administration, and cloud architecture concepts without cost.

No. The Always Free resources persist indefinitely beyond the initial thirty-day trial. Only paid promotional credits expire. Your ARM compute instances, storage volumes, and networking components remain active as long as you use them at least once every seven days.

Set up budget alerts in the OCI Console billing section immediately. Create a zero-spend budget rule that triggers email notifications if any paid service activates. Disable automatic upgrades and verify resource shapes match Always Free limits before launching instances or provisioning storage.

Yes. Install Docker CE on your Always Free ARM or AMD instance. The four ARM cores and 24GB RAM handle multiple containers comfortably. Use podman as an alternative for rootless execution. Monitor memory usage since swap space is limited on free tier instances.

Oracle Linux 8 and Ubuntu 22.04 LTS have optimized images with preconfigured cloud-init and kernel modules. AlmaLinux 9 and Rocky Linux 9 also work well. Avoid minimal images lacking OCI guest tools, as they cause metadata service failures and network configuration issues during boot.

Idle Always Free compute instances stop automatically after seven consecutive days of inactivity. Restart via console or CLI. Prevent this by scheduling periodic SSH connections or cron jobs. Check audit logs to confirm whether manual termination or idle reclamation caused the stopped state.

Yes. The ARM instance runs PHP 8.3, Nginx, and MySQL efficiently. Configure OPcache and use Redis from Always Free caching tiers. Deploy with Laravel Forge or manually via Composer. The 24GB RAM supports local testing, queue workers, and database operations simultaneously.

Oracle offers permanent ARM compute with 24GB RAM versus AWS twelve-month t2.micro limitations. Oracle includes 200GB persistent storage compared to AWS 30GB EBS. For sustained DevOps practice beyond one year, Oracle provides significantly more usable resources without requiring credit card renewal or account recreation.

Yes. Allocate the four ARM cores across up to four separate instances or use two AMD micro instances. Splitting resources enables multi-node Kubernetes testing or isolated environments. Total allocation cannot exceed four OCPUs and 24GB RAM across all ARM instances combined.

Generate ed25519 keys locally and upload the public key during instance creation. Disable password authentication in sshd_config. Restrict ingress rules in the default security list to your IP only. Store private keys outside the instance and never embed credentials in user-data scripts.

Instance creation fails with an out-of-capacity error rather than charging overages. Existing Always Free resources continue running. To scale temporarily, upgrade to pay-as-you-go billing explicitly. Monitor usage dashboards weekly to ensure configurations stay within free allocations and avoid unexpected service interruptions.

Yes. Open ports 80 and 443 in the VCN security list and OS firewall. Use Caddy or Nginx with Let's Encrypt certificates. Point DNS to the public IP. Remember that Always Free instances lack SLAs and may experience maintenance downtime unsuitable for production traffic.

No automated backups exist for Always Free resources. Create manual boot volume clones periodically through the console. Export critical data to Object Storage free tier monthly. Implement custom backup scripts using rclone or restic since losing an idle instance means losing all uncommitted data permanently.

Terminate all instances and delete associated boot volumes, VCNs, and subnets through the console. Wait ten minutes for resource release. Recreate infrastructure using Terraform or Ansible to ensure clean state. This removes accumulated misconfigurations and restores baseline conditions for fresh learning exercises.

Yes. Self-host GitLab Runner or Gitea Actions on ARM instances. Four cores handle parallel jobs adequately for personal projects. Register runners as non-root users with Docker socket access. Expect slower performance than paid tiers but sufficient throughput for learning pipeline configuration and automation workflows.