A practical cloud security baseline checklist helps teams review AWS, Azure, and Google Cloud environments consistently, assign remediation work, and preserve evidence for cybersecurity compliance. Use this guide for onboarding, quarterly reviews, major architecture changes, and pre-audit preparation—not as a substitute for a formal audit or provider-specific benchmark.
Overview
Cloud configuration changes quickly. A new account, public storage setting, overly broad role, missing log destination, or untested backup can alter your risk profile without changing your application code. A baseline review creates a repeatable way to identify those changes and compare the environment with your organization’s security and compliance expectations.
The baseline should combine three perspectives:
- Configuration: Is the control enabled and correctly configured?
- Exposure: Could a mistake expose data, credentials, workloads, or management interfaces?
- Evidence: Can the team show what was checked, when it was checked, who reviewed it, and how exceptions were handled?
Provider benchmarks and security services can support the review, but they do not define your complete risk program. Map each check to an internal policy, system owner, data classification, and relevant framework requirement. For example, a review may support SOC 2, ISO 27001, GDPR security obligations, or sector-specific requirements, while still requiring separate assessments of governance, privacy operations, contracts, and incident response.
Keep the checklist in a version-controlled location. Record the review date, cloud account or subscription, environment, reviewer, scope, benchmark or policy version, finding severity, owner, due date, and evidence location. This turns a point-in-time scan into an audit evidence workflow.
What to track
Use the following control areas as a provider-neutral cloud security baseline checklist. Adapt the exact setting to the services and architecture you operate.
1. Identity and privileged access
- Confirm that human users use individually assigned identities rather than shared administrator accounts.
- Require strong authentication for privileged access and review emergency or break-glass accounts.
- Remove inactive identities, unused access keys, stale service accounts, and permissions that exceed the job requirement.
- Separate production access from development access and record approved administrative roles.
- Review privileged access on a recurring schedule. The cloud privileged access review checklist can help structure certifications.
Implementation examples include IAM and Access Analyzer in AWS, Microsoft Entra ID and Azure role assignments, and Google Cloud IAM with organization or project-level policies. Names and console paths change, so verify the current provider documentation before publishing an internal procedure.
2. Network exposure
- Inventory internet-facing load balancers, virtual machines, containers, databases, storage endpoints, and management interfaces.
- Restrict inbound administration to approved paths, such as a controlled bastion, private connectivity, or an authorized corporate network.
- Review firewall rules, security groups, network security groups, routes, and egress controls for broad or unnecessary access.
- Separate production, non-production, and sensitive workloads where the risk assessment requires it.
- Document intentional public services, their owner, business purpose, and compensating controls.
3. Encryption and key management
- Confirm encryption is enabled for storage, databases, backups, message queues, and other sensitive data services.
- Identify who controls encryption keys, how access is restricted, and how key rotation or replacement is managed.
- Check that transport encryption is used for administrative and application traffic where appropriate.
- Record exceptions for legacy integrations or services that cannot meet the baseline.
Review AWS KMS, Azure Key Vault, and Google Cloud KMS configurations alongside application-level encryption requirements. Encryption alone does not resolve excessive identity access or an exposed endpoint.
4. Logging, monitoring, and alerting
- Enable management-plane and relevant data-plane logs for accounts, subscriptions, projects, and critical services.
- Send logs to a protected, centralized destination with restricted deletion and alteration rights.
- Define retention periods that match security, legal, contractual, and privacy requirements.
- Confirm that alerts exist for high-risk events such as privilege changes, logging disablement, public access changes, suspicious authentication, and key modifications.
- Test whether someone reviews alerts and records investigation outcomes.
Typical implementation choices include CloudTrail and related monitoring in AWS, Activity Logs and Azure Monitor in Azure, and Cloud Audit Logs with relevant monitoring services in Google Cloud. Capture configuration screenshots or exported settings only when they are sufficient; machine-readable reports and immutable logs are often more useful evidence.
5. Secrets and application credentials
- Locate secrets in source repositories, build logs, images, configuration files, tickets, and collaboration tools.
- Move credentials into an approved secrets-management service and restrict retrieval to the workloads that need it.
- Rotate exposed or long-lived credentials, including credentials found during scans.
- Use short-lived credentials or workload identities where the platform and application support them.
6. Storage permissions and data protection
- Scan object storage, file shares, snapshots, and databases for public or cross-account access.
- Confirm that access policies match data classification and documented business use.
- Review sharing links, temporary URLs, replication destinations, and export paths.
- Connect storage findings to retention and deletion rules. A data retention and deletion policy helps clarify when data should be kept, archived, or removed.
7. Backup and recovery
- List critical workloads and verify that backup coverage matches recovery requirements.
- Check backup schedules, retention, cross-region or cross-account protection, and access restrictions.
- Test restoration for representative systems and retain the result, including failures and corrective actions.
- Ensure backups are not automatically exposed through the same identity or network path as production.
8. Vulnerability management and monitoring coverage
- Maintain an inventory of virtual machines, container images, managed services, operating systems, and application dependencies.
- Track critical vulnerabilities, affected assets, compensating controls, owner, target date, and verification evidence.
- Confirm that monitoring covers new accounts, regions, projects, subscriptions, and ephemeral workloads.
- Review unsupported software and platform versions during architecture changes and procurement decisions.
Security services such as AWS Config or Security Hub, Microsoft Defender for Cloud or Azure Policy, and Google Cloud Security Command Center can help identify drift. Treat their findings as inputs to a risk-based process, not automatic proof that every control is effective.
Provider comparison at a glance
| Review area | AWS examples | Azure examples | Google Cloud examples |
|---|---|---|---|
| Configuration and policy | AWS Config, Organizations policies | Azure Policy, management groups | Organization policies, Security Command Center |
| Identity | IAM, IAM Access Analyzer | Microsoft Entra ID, Azure RBAC | Cloud IAM, organization controls |
| Logging | CloudTrail and centralized log storage | Activity Logs, Azure Monitor | Cloud Audit Logs, Cloud Logging |
| Secrets and keys | Secrets Manager, KMS | Key Vault | Secret Manager, Cloud KMS |
| Threat and vulnerability visibility | Security Hub and provider security services | Defender for Cloud | Security Command Center and provider security services |
Service availability, names, and recommended settings vary by region, account type, workload, and platform release. Validate each implementation against current provider documentation and your approved architecture.
Cadence and checkpoints
A useful cadence combines continuous detection with scheduled human review:
- Continuously or daily: collect configuration drift, public exposure, identity, vulnerability, and logging alerts.
- Monthly: review open high-severity findings, new assets, exposed services, failed backups, and unowned resources.
- Quarterly: perform the full baseline review across accounts, subscriptions, projects, regions, and environments. Reconfirm privileged access and exceptions.
- After a major change: review new services, network paths, identities, data stores, backup coverage, logging, and privacy implications.
- Before an audit or customer review: verify that findings are closed or formally accepted and that evidence links remain accessible.
Assign one accountable owner for each finding, even when several teams contribute to remediation. Set a target date based on severity and exposure. A practical severity model is: critical for active or likely compromise of sensitive systems; high for significant exposure or loss of a key control; medium for material weakness with mitigating controls; and low for hygiene or documentation gaps. Adjust these definitions to your risk methodology.
How to interpret changes
Do not measure success only by the number of findings. Interpret changes in context:
- More findings may indicate better visibility. New accounts, regions, scanners, or data sources can increase reported issues without a corresponding increase in underlying risk.
- Fewer findings may indicate closure—or weaker monitoring. Confirm that the scanner ran, its scope did not shrink, and logs are still arriving.
- Repeated findings indicate a process problem. Consider preventive guardrails, approved templates, policy-as-code, or automated remediation rather than repeatedly fixing individual resources.
- Exceptions are not failures when they are controlled. Document the rationale, affected asset, risk owner, compensating control, expiration date, and review decision.
For every closed finding, retain before-and-after evidence and a verification result. For accepted risk, retain approval and a future review date. Link the control to relevant framework requirements, contracts, or customer commitments without claiming that one configuration check proves full compliance. For broader framework planning, compare the control expectations in the SOC 2 and ISO 27001 cloud compliance guide.
When to revisit
Review this checklist at least quarterly and update it after major cloud-platform releases, benchmark revisions, new services, or changes to your threat model. Revisit it immediately when you add a cloud account, subscription, or project; launch a sensitive workload; change identity architecture; migrate data; introduce a new vendor; or experience a security incident.
At each review, record the baseline version and compare the current asset inventory with the previous period. Ask:
- Did the scope change, and are all environments still included?
- Were any provider services, benchmark recommendations, or internal policies updated?
- Did new data types, processing purposes, or geographic locations change the privacy risk?
- Are high-risk exceptions expired, still owned, and still justified?
- Can the team retrieve evidence for configuration, review, remediation, and approval?
Start with a small, controlled scope if your environment is large: production accounts and sensitive data stores first, followed by non-production and lower-risk resources. Create a dated remediation backlog, schedule verification, and communicate unresolved risks to the accountable service owner. This makes a cloud security baseline a living control for cloud compliance—one that supports ongoing cybersecurity compliance and audit readiness without pretending that configuration review alone completes the job.