Skip to content

Patch Management and Vulnerability Response

Last reviewed: August 2026

Operating cloud infrastructure involves two recurring security tasks.

  1. Periodic patching — Regularly applying security updates to the OS, runtime, and middleware
  2. Vulnerability response — When a CVE is disclosed, assessing the scope of impact and applying an urgent patch or mitigation

These are separate processes, but since tools and pipelines overlap, they’re covered together here.

Vendor Service Target Characteristics
AWS Systems Manager Patch Manager EC2 (Linux/Windows) Defines patch baselines, schedules maintenance windows, compliance reporting
Azure Azure Update Manager VM (Linux/Windows), Arc-connected servers Agentless assessment, scheduled patching, pre/post scripts
Google Cloud OS Patch Management Compute Engine (Linux/Windows) Patch Jobs + Patch Deployments
OCI OS Management Hub Compute (Oracle Linux/Windows) Package management, scheduled jobs, compliance reporting
graph LR
    A[Patch scan<br/>assessment] --> B[Baseline<br/>approve/exclude] --> C[Test environment<br/>apply+validate] --> D[Production<br/>rolling apply]
  1. Scan/assess — Check the list of missing patches on current instances
  2. Define baseline — Classify which patch categories to auto-approve (Critical, Security, etc.) and set an exclusion list
  3. Test application — Apply to a non-production environment first to verify compatibility
  4. Production rollout — Apply sequentially by group within a maintenance window. Halt on failure
Strategy Description Suitable environment
Immutable Infrastructure Replace instances with a patched new AMI/image Containers, Auto Scaling groups, serverless
In-place Patching Apply patches directly to a running instance Stateful servers (DB, legacy apps)
Blue-Green Prepare a patched new environment, then switch traffic When minimizing downtime is required

These services overlap with the CWPP area of Security Posture Management. Here they’re covered from a patch-response perspective.

Vendor Service Scan target Characteristics
AWS Amazon Inspector EC2, ECR images, Lambda Agentless automatic scanning. CVE + network reachability analysis
Azure Microsoft Defender for Cloud VM, containers, App Service, DB Integrated CSPM + CWPP. Vulnerability assessment + security recommendations
Google Cloud Security Command Center + Artifact Analysis Compute Engine, GKE, Artifact Registry Integrated vulnerability + misconfiguration + threat detection
OCI Vulnerability Scanning Compute, Container Registry Agent-based host scanning + container image scanning

In container environments, detecting vulnerabilities at image build time is essential.

Vendor Registry scanning CI/CD integration
AWS ECR image scanning (integrated with Inspector) Automatic scanning in CodeBuild/CodePipeline
Azure Defender for Containers (ACR scanning) Azure DevOps/GitHub Actions integration
Google Cloud Artifact Analysis (Artifact Registry) Automatic scan on Cloud Build triggers
OCI Container Registry scanning Integrated with DevOps service pipelines
graph LR
    A[CVE disclosed] --> B[Impact assessment] --> C[Priority determined] --> D[Mitigate·patch] --> E[Verify] --> F[Deploy]
Stage Activity Example tools
Detection The vulnerability scanner matches CVEs Inspector, Defender, SCC
Impact assessment CVSS score + actual exposure (network reachability, whether used at runtime) Inspector network reachability, Defender attack path analysis
Prioritization Critical + externally exposed = respond immediately. Low + internal only = next regular patch cycle CVSS + EPSS + business impact
Mitigation Temporary measures before patching (WAF rule, network isolation, disabling the feature) WAF, Security Group, feature flag
Apply patch Regular patch process or an urgent hotfix Patch Manager, image rebuild
Verify Rescan for vulnerabilities after patching + confirm normal service operation Rescan + smoke test

Emergency Vulnerability (Zero-Day) Response

Section titled “Emergency Vulnerability (Zero-Day) Response”

The response framework for urgent situations that can’t wait for the regular patch cycle.

Step Activity
1. Receive alert Monitor vendor security bulletins, CVE feeds, security news
2. Identify scope Immediately query the list of resources using the vulnerable package/version (leveraging SBOM)
3. Immediate mitigation Network isolation, WAF virtual patching, disabling the vulnerable feature
4. Emergency patch Apply the vendor patch release immediately (reduced testing may be acceptable, but always prepare a rollback plan)
5. Post-verification Rescan the entire environment, check for indicators of compromise (IOC)
  • Golden image pipeline — Periodically build patched base images, and deploy all instances/containers from them
  • Compliance dashboard — Real-time monitoring of the proportion of unpatched instances (target: 95%+ compliance)
  • SLA-based patch deadlines — Critical: within 72 hours, High: within 7 days, Medium: within 30 days (adjust per organizational policy)
  • Automatic ticket creation — Automatically create a Jira/ServiceNow ticket when a vulnerability is detected, for tracking
  • Regular reporting — Monthly patch compliance rate, mean time to patch (MTTP), trend of unresolved vulnerabilities
  • Applying patches directly to production without testing — Compatibility issues cause service incidents. Always validate in a non-production environment first.
  • Deciding priority by CVSS score alone — You need to also consider actual exploit likelihood (EPSS) and network exposure. A High CVE on an externally exposed server can be more urgent than a Critical CVE on an internal-only server.
  • Not maintaining an SBOM, causing days of delay in identifying impact scope during an emergency CVE — Enable automatic SBOM generation in Inspector/Artifact Analysis.
  • Is the proportion of unpatched instances monitored in real time (target: 95%+ compliance)?
  • Is a patch SLA defined for Critical vulnerabilities (e.g., within 72 hours)?
  • Is there a golden image pipeline that periodically rebuilds container base images?