Skip to main content

Security Best Practices for CloudWatch Logs

Securing your Amazon CloudWatch Logs is essential for maintaining compliance, protecting sensitive data, and ensuring proper audit trails. This guide provides comprehensive best practices for implementing robust permission controls and security policies around your log groups, including the critical deletion protection feature.

Introduction​

Amazon CloudWatch Logs enables you to centralize logs from your systems, applications, and AWS services into a single, highly scalable service (What is Amazon CloudWatch Logs?). However, without proper security controls, log data can become a vulnerability rather than an asset. This guide focuses on implementing least-privilege access, encryption, resource-based policies, deletion protection, and comprehensive auditing to keep your log groups secure and compliant.

Why This Matters​

Security Implications​

Log data often contains sensitive information including user activities, system configurations, API calls, and potentially personally identifiable information (PII). Unauthorized access to logs can expose critical security details about your infrastructure, application behavior, and business operations. Additionally, accidental or malicious deletion of log groups can result in loss of critical audit trails and compliance violations.

Compliance Requirements​

Many regulatory frameworks require specific controls around log data including access restrictions, encryption at rest and in transit, retention policies, deletion protection, and audit trails. Proper permission management and deletion protection are fundamental to meeting these requirements.

Operational Excellence​

Well-structured permissions enable teams to access the logs they need while preventing unwanted modifications and deletions. This balance supports both security and operational efficiency while maintaining data integrity.

Security Best Practices​

CloudWatch Logs security operates through multiple layers of access control, deletion protection, and encryption mechanisms that work together to protect your log data. Implementing comprehensive security requires a multi-layered approach combining IAM policies, deletion protection, encryption, resource policies, and continuous monitoring.

1. CloudWatch Logs Hierarchy and Security Boundaries​

Understanding CloudWatch Logs architecture is fundamental to implementing effective security controls. Proper log organization and hierarchy design form the foundation for all other security measures.

CloudWatch Logs uses a two-level hierarchy that directly impacts security controls (Working with log groups and log streams):

  • Log Groups: Top-level containers that define retention policies, encryption settings, access permissions, and deletion protection. Each log group acts as a security boundary with its own IAM policies and KMS encryption keys
  • Log Streams: Individual sequences of log events within a log group, typically representing a single source (like an EC2 instance, Lambda function, or application process). Log streams inherit security settings from their parent log group but can be individually targeted in IAM policies for granular access control

Security-Driven Log Group Design​

Design your log group structure to align with security requirements and access patterns (CloudWatch Logs permissions reference):

  • Application Separation: Create distinct log groups for different applications, especially when handling sensitive data, to enable granular IAM policies and prevent cross-application log access
  • Environment Isolation: Use separate log groups for production, staging, and development environments to enforce different access controls and retention policies
  • Data Classification: Group logs by sensitivity level (public, internal, confidential, restricted) to apply appropriate encryption, access controls, and retention policies
  • Compliance Boundaries: Create dedicated log groups for audit logs, security logs, and compliance-related data that require special handling and longer retention periods

Granular Access Control with Log Streams​

While log groups provide the primary security boundary, log streams enable fine-grained access patterns (Actions, resources, and condition keys for CloudWatch Logs):

  • Instance-Level Access: Use log stream names in IAM policies to grant users access only to logs from specific EC2 instances or containers
  • Time-Based Access: Implement policies that restrict access to log streams based on creation time or naming patterns
  • Service-Specific Streams: Allow applications to write only to their designated log streams while preventing access to other streams in the same log group
  • Audit Trail Integrity: Use log stream immutability (once created, log events cannot be modified) as part of your audit and compliance strategy

2. Identity-Based Policies (IAM Policies)​

  • Use IAM policies to control who can create, read, and manage log groups and log streams (Using identity-based policies for CloudWatch Logs)

    • Apply Least-Privilege Principles: Create customer-managed policies that restrict access to specific log groups based on your organizational needs
    • Use Specific Resource ARNs: Always specify explicit log group ARNs in your IAM policies rather than using wildcards (*). This prevents privilege escalation and ensures users can only access intended log groups
    • Separate Administrative and Operational Permissions: Create distinct policies for different permission levels for example, read-only access for analysts, write permissions for applications, and administrative permissions for infrastructure teams. Never combine these in a single overly permissive policy
    • Explicitly Deny Deletion Operations: For critical log groups, implement explicit deny statements for deletion operations to provide additional protection beyond deletion protection

  • For Lambda functions logging to CloudWatch, ensure IAM roles include minimum required permissions: logs:CreateLogGroup, logs:CreateLogStream, and logs:PutLogEvents (Lambda execution role)

  • Implement MFA for privileged accounts that can modify or delete log groups (Overview of managing access permissions)

  • Implement Tag-Based Access Control: Use resource tags on log groups combined with IAM condition keys (aws:ResourceTag) to dynamically control access based on attributes like environment (production/development), team ownership, or data classification level (Actions, resources, and condition keys for CloudWatch Logs)

3. Deletion Protection for Critical Log Groups​

Deletion protection is a critical security feature introduced by Amazon CloudWatch Logs that prevents accidental or malicious deletion of log groups and their associated log streams. When enabled, deletion protection blocks all deletion operations until it is explicitly disabled, helping protect critical operational and compliance data. This feature is particularly valuable for preserving audit logs, compliance records, and production application logs that must be retained for troubleshooting, analysis, and regulatory requirements. (Protecting log groups from deletion)

Key Characteristics:​

  • Preventive Control: Acts as a preventive security control that stops deletion attempts before they occur
  • Explicit Disable Required: Must be explicitly disabled before any deletion operation can proceed
  • Applies to Log Streams: Protects both the log group and all log streams within it
  • No Performance Impact: Does not affect log ingestion, querying, or other operations
  • Audit Trail: All changes to deletion protection status are logged in CloudTrail

Critical Use Cases - When to Enable Deletion Protection:​

  • Audit Logs: All audit logs should have deletion protection enabled to maintain compliance and prevent tampering with audit trails
  • Security Logs: Security-related logs including AWS CloudTrail, VPC Flow Logs, and application security logs
  • Compliance Logs: Any logs required for regulatory compliance
  • Production Application Logs: Production logs needed for troubleshooting and incident response
  • Long-Term Retention Logs: Any logs with retention requirements exceeding 1 year

Implementation Best Practices:​

  • Enable on Critical Log Groups: Enable deletion protection during log group creation or on existing log groups for all critical logs. This is your first line of defense against accidental or malicious deletion
  • Automate Deployment: Use Infrastructure as Code (IaC) tools like AWS CloudFormation, AWS CDK, or Terraform to automatically enable deletion protection when creating new log groups. This ensures consistent security posture across your environment
  • Document Procedures: Create clear documentation and runbooks for when and how to disable deletion protection. This should include approval workflows, justification requirements, and re-enablement procedures to ensure protection is only temporarily disabled when absolutely necessary
  • Monitor Changes: Create CloudWatch alarms and metric filters to detect when deletion protection is disabled on critical log groups. Alert security teams immediately when this occurs to investigate whether the change was authorized
  • Defense in Depth: Use deletion protection in combination with IAM policies that explicitly deny deletion operations. This provides defense in depth - even if deletion protection is disabled, IAM policies can still prevent unauthorized deletion

Combine Multiple Protection Layers:​

4. Encryption with Customer-Managed KMS Keys​

Implement customer-managed KMS keys for sensitive log groups to maintain full control over encryption keys, enable key rotation, and create detailed audit trails of key usage.

Encryption Architecture​

Encryption Best Practices:​

  • Enable KMS Encryption for Sensitive Logs: For log groups containing sensitive data, associate a customer-managed KMS key. This provides enhanced control through key policies, enables key rotation, and creates detailed CloudTrail logs of all encryption/decryption operations
  • Configure Proper KMS Key Policies: Your KMS key policy must grant CloudWatch Logs service principal (logs.amazonaws.com) permissions to use the key for encryption and decryption. Include conditions that restrict usage to specific log groups and AWS accounts
  • Implement Key Rotation: Enable automatic key rotation for your KMS keys used with CloudWatch Logs. AWS automatically rotates customer-managed keys annually while maintaining access to data encrypted with previous key versions
  • Monitor KMS Key Usage: Use CloudTrail to monitor all KMS API calls related to your log encryption keys. Set up CloudWatch alarms to alert on unusual patterns like excessive decrypt operations or unauthorized key access attempts

5. Data Protection Policies​

CloudWatch Logs data protection is a feature that helps you discover, protect, and audit sensitive data in your log groups using machine learning and pattern matching (Protecting sensitive log data with masking). This feature automatically scans log events for sensitive information such as personally identifiable information (PII), credentials, and financial data, then either audits or masks the data based on your configuration. Data protection policies work in real-time as log events are ingested, providing immediate protection without requiring changes to your applications or log sources.

  • Configure Data Protection Policies: Implement CloudWatch Logs data protection policies to automatically detect and mask sensitive information using managed data identifiers (Personally identifiable information (PII))
  • Choose Protection Operations: Configure either audit operations (to monitor and report sensitive data) or de-identify operations (to mask sensitive data in real-time) based on your security requirements (CloudWatch Logs managed data identifiers for sensitive data types)
  • Comprehensive Data Coverage: Protect multiple categories of sensitive data including:
    • Credentials: AWS Secret Keys, SSH Private Keys, PGP Private Keys, PKCS Private Keys
    • Financial: Credit card numbers, bank account numbers, security codes
    • Personal: Email addresses, names, addresses, IP addresses, vehicle identification numbers
    • Regional-Specific: Country-specific identifiers like driver's licenses, tax IDs, postal codes
  • Keyword Proximity Detection: Leverage advanced detection that scans for keywords within 30 characters of sensitive data patterns to reduce false positives
  • Global Coverage: Data protection policies work regardless of log group geolocation, with support for region-specific data identifiers using ISO country codes
  • Integration with Amazon Macie: Use Amazon Macie in conjunction with CloudWatch Logs for enhanced sensitive data discovery and classification across your AWS environment (AWS Blog: How Amazon CloudWatch Logs Data Protection can help detect and protect sensitive log data)

6. Log Retention and Lifecycle Management​

Log retention in CloudWatch Logs controls how long log events are stored before being automatically deleted, helping you balance compliance requirements with storage costs (Change log data retention in CloudWatch Logs). By default, CloudWatch Logs stores log data indefinitely, but you can configure retention periods ranging from 1 day to 10 years at the log group level. Proper lifecycle management ensures that sensitive data is retained for the appropriate duration based on regulatory requirements, operational needs, and cost optimization goals, while automatically purging data when it's no longer needed.

  • Configure Retention Periods: Set appropriate retention periods for log groups based on compliance requirements and operational needs. By default, log data is stored indefinitely, but you can configure retention from 1 day to 10 years (Change log data retention in CloudWatch Logs)
  • Apply Data Classification-Based Policies: Implement different retention policies based on data sensitivity and classification:
    • Critical/Audit Logs: Long-term retention (7+ years) for compliance requirements
    • Security Logs: Extended retention (1-3 years) for forensic analysis
    • Application Logs: Medium-term retention (30-90 days) for troubleshooting
    • Debug/Development Logs: Short-term retention (1-7 days) for cost optimization
  • Cost Optimization: Regularly review and adjust retention periods to balance compliance needs with storage costs. Older log data is automatically deleted when retention periods expire
  • Tagging for Lifecycle Management: Use consistent tagging strategies to categorize log groups by environment, application, data classification, and retention requirements for automated policy application (Tag log groups in Amazon CloudWatch Logs)
  • Integration with Centralization: When using log centralization, ensure retention policies are consistently applied across source and destination accounts to maintain compliance requirements

7. Resource-Based Policies for Log Destinations​

Resource-based policies in CloudWatch Logs are specifically used for destinations to enable cross-account subscriptions (Cross-account cross-Region subscriptions). Unlike log groups which use identity-based IAM policies, destinations support resource-based policies that specify which external AWS accounts can subscribe their log groups to your destination resources like Kinesis Data Streams, Kinesis Data Firehose, or Lambda functions (Overview of managing access permissions).

What are Destinations and When to Use Resource-Based Policies:​

  • Log Destinations: Destinations are CloudWatch Logs resources that represent AWS services (Kinesis Data Streams, Kinesis Data Firehose, Lambda functions) that can receive log data from subscription filters
  • Cross-Account Log Streaming: Use resource-based policies on destinations when you want to allow other AWS accounts to stream their log data to your centralized processing infrastructure
  • Centralized Log Processing: Enable scenarios where multiple accounts send logs to a central account's Kinesis stream or Firehose for unified analysis, security monitoring, or compliance processing
  • Third-Party Integration: Allow partner accounts or service providers to send log data to your processing systems while maintaining strict access controls

Resource-Based Policy Best Practices for Destinations:​

  • Specify Exact Source Accounts: In destination policies, explicitly specify the AWS account IDs that are allowed to create subscriptions. Never use wildcards (*) for account IDs (Step 1: Create a destination)
  • Use Least-Privilege Access: Grant only the minimum permissions required - typically just logs:PutSubscriptionFilter for the specific destination ARN
  • Implement Condition Keys: Use IAM condition keys to add additional security layers such as source IP restrictions, time-based access, or MFA requirements
  • Regular Policy Audits: Periodically review destination policies to ensure they still reflect current requirements. Remove access for decommissioned accounts and tighten overly permissive policies
  • Monitor Subscription Activity: Use CloudTrail to monitor PutSubscriptionFilter and DeleteSubscriptionFilter API calls to track which accounts are creating or removing subscriptions to your destinations

8. Log Centralization with AWS Organizations​

Log centralization is an AWS Organizations feature that automatically replicates log data from multiple member accounts and AWS Regions into a centralized account using cross-account and cross-region centralization rules (Cross-account cross-Region log centralization). This capability streamlines log consolidation for improved centralized monitoring, analysis, and compliance across your entire AWS infrastructure without requiring resource-based policies or cross-account IAM roles. The feature provides configuration flexibility to meet operational and security requirements, including backup region configuration and full control over encryption behavior for log groups copied from source accounts.

AWS Security Reference Architecture Alignment​

Following AWS Security Reference Architecture (AWS SRA) best practices, CloudWatch Logs centralization should align with your overall security account structure (AWS Security Reference Architecture):

  • Designate Log Archive Account as Delegated Administrator: Configure your dedicated Log Archive account as the CloudWatch delegated administrator for your AWS Organization. This account should be dedicated to ingesting and archiving all security-related logs and provides immutable storage with controlled access mechanisms
  • Centralize to Security OU: Direct all CloudWatch Logs centralization rules to replicate logs into the Log Archive account within your Security Organizational Unit (OU), ensuring separation from production workloads and consistent security controls
  • Integrate with Existing Log Infrastructure: Leverage the Log Archive account's existing security infrastructure, including customer-managed KMS encryption keys, IAM access control patterns, VPC endpoints, and monitoring frameworks already established for other security logs, while maintaining CloudWatch Logs as the primary storage for centralized log groups
  • Implement Defense in Depth: Apply the same security principles used for other critical logs - least privilege access, encryption with customer-managed KMS keys, deletion protection for immutability, and comprehensive monitoring

Log Centralization Best Practices:​

  • Establish Organizational Structure: Designate the Log Archive account as the CloudWatch delegated administrator and create centralization rules to replicate logs from all member accounts across your organization
  • Apply Consistent Security Controls: Implement uniform security policies across all centralized log groups including:
    • Encryption: Use the same customer-managed KMS keys already established in the Log Archive account for other security logs
    • Access Policies: Apply consistent IAM policies that align with your existing log access controls and separation of duties
    • Retention: Configure retention policies that meet your compliance requirements and integrate with existing log lifecycle management
  • Monitor Centralization Health: Use CloudWatch metrics and console monitoring to track the health status of centralization rules, identify replication issues, and ensure continuous log flow from all member accounts
  • Integrate with Existing Log Sources: Coordinate CloudWatch Logs centralization with other log sources already flowing to the Log Archive account (CloudTrail, VPC Flow Logs, GuardDuty findings) for unified log management and analysis
  • Configure Multiple Centralization Rules for Data Residency: Use multiple centralization rules to address data residency and compliance requirements:
    • Regional Data Residency: Create separate centralization rules for different regions to ensure log data remains within specific geographic boundaries required by regulations like GDPR, data sovereignty laws, or organizational policies
    • Compliance-Based Segregation: Configure distinct centralization rules for different types of sensitive data (financial, healthcare, personal data) to meet industry-specific compliance requirements
    • Multi-Region Backup Strategy: Implement centralization rules that replicate critical logs to multiple regions for disaster recovery while respecting data residency constraints
    • Selective Log Routing: Use centralization rule filters to route specific log types to appropriate destination accounts and regions based on data classification and residency requirements
  • Monitor Centralization Health: Use CloudWatch metrics and console monitoring to track the health status of centralization rules and identify replication issues
  • Implement Centralized Security Controls: Apply consistent security policies, encryption settings, and access controls across all centralized log groups to maintain uniform security posture

9. VPC Endpoints for CloudWatch Logs​

Use VPC endpoints to establish private connectivity between your VPC and CloudWatch Logs, keeping log traffic within the AWS network and enhancing security through network isolation.

  • Enable Private Connectivity: Use interface VPC endpoints to send logs to CloudWatch Logs without traversing the internet (Using CloudWatch Logs with interface VPC endpoints)
  • Multiple Endpoint Support: CloudWatch Logs requires two VPC endpoints:
    • com.amazonaws.region.logs for standard CloudWatch Logs APIs
    • com.amazonaws.region.stream-logs for streaming APIs like StartLiveTail and GetLogObject
  • FIPS Endpoint Support: Use FIPS-compliant endpoints (logs-fips and stream-logs-fips) when required for compliance
  • Implement VPC Endpoint Policies: Use endpoint policies to restrict CloudWatch Logs actions through the VPC endpoint, such as allowing only log creation and ingestion while preventing administrative operations
  • Leverage VPC Context Keys: Use aws:SourceVpc and aws:SourceVpce condition keys in IAM policies to ensure CloudWatch Logs is only accessible through specific VPC endpoints
  • Network Perimeter Security: Logs containing sensitive security and audit information remain within your controlled network perimeter, preventing accidental data exfiltration through public endpoints

10. Monitoring and Auditing​

Enable Comprehensive Logging:​

  • Enable CloudTrail Logging: Ensure CloudTrail is enabled in all regions and configured to log CloudWatch Logs API calls. Configure CloudTrail to send events directly to CloudWatch Logs log groups for monitoring and analysis (Sending events to CloudWatch Logs)
  • Configure CloudWatch alarms to detect unauthorized access attempts or unusual patterns (Using Amazon CloudWatch alarms)
  • Implement centralized logging across multiple accounts using Organizations (Cross-account log data sharing in CloudWatch Logs)
  • Maintain immutable audit trails by preventing log deletion through IAM policies
  • Create separate log groups for applications with different sensitivity levels

Monitoring Best Practices:​

  • Monitor Log Ingestion Metrics: Use built-in CloudWatch metrics to track log ingestion patterns and detect anomalies (CloudWatch Logs metrics):

    • IncomingLogEvents: Monitor the number of log events ingested to detect unusual spikes or drops that might indicate security incidents, application issues, or unauthorized log sources
    • IncomingBytes: Track the volume of log data being ingested to identify potential data exfiltration attempts or denial-of-service attacks through excessive logging
    • DeliveryErrors: Monitor failed log deliveries which could indicate tampering with log destinations or infrastructure issues affecting audit trails
    • Set Baseline Thresholds: Establish normal ingestion patterns and create CloudWatch alarms for deviations that exceed acceptable variance ranges

  • Leverage CloudWatch Contributor Insights for Anomaly Detection: Use Contributor Insights to analyze log data patterns and identify unusual activity (Using CloudWatch Contributor Insights to Analyze High-Cardinality Data):

    • Top Talkers Analysis: Identify the highest volume log contributors (IP addresses, user agents, API endpoints) to detect potential abuse, bot activity, or security threats
    • Error Pattern Detection: Analyze error logs to identify unusual error patterns, failed authentication attempts, or suspicious access patterns that might indicate security incidents
    • Resource Usage Monitoring: Track which users, applications, or services are generating the most log data to identify potential misuse or unauthorized activity
    • Time-Based Analysis: Use time-series analysis to detect unusual activity patterns during off-hours or unexpected traffic spikes that could indicate compromise

  • Create Security Event Alarms: Set up metric filters and alarms to detect suspicious activities such as unauthorized log group deletions, deletion protection changes, permission changes, encryption key disassociation, or unusual query patterns (Creating metric filters)

  • Set up CloudWatch dashboards to visualize security metrics and access patterns (Using Amazon CloudWatch dashboards)

  • Use CloudWatch Logs Insights for advanced log analysis and anomaly detection (Analyzing log data with CloudWatch Logs Insights)

  • Use AWS Config to monitor CloudWatch Logs configuration changes (Monitoring AWS Config with Amazon EventBridge)

  • Implement AWS Security Hub to aggregate and prioritize security findings related to logging (AWS Security Hub User Guide)

  • Monitor Failed Access Attempts: Create metric filters to track failed API calls to CloudWatch Logs (AccessDenied errors). Alert security teams when patterns suggest attempted unauthorized access or privilege escalation

  • Implement Deletion Protection Monitoring: Use IAM policies with explicit deny statements and CloudWatch alarms to monitor attempts to disable deletion protection or delete protected log groups. Consider using AWS Organizations Service Control Policies (SCPs) for organization-wide protection

Conclusion​

Securing Amazon CloudWatch Logs requires a comprehensive, multi-layered approach that combines identity-based policies, deletion protection, encryption, data protection policies, and continuous monitoring to protect your critical log data. By implementing these security best practices—from least-privilege IAM policies and deletion protection to VPC endpoints and automated sensitive data detection—you create a robust defense against both accidental and malicious threats to your log infrastructure. These controls not only protect sensitive operational and compliance data but also ensure your organization meets regulatory requirements while maintaining the operational visibility needed for effective monitoring and troubleshooting. Proper CloudWatch Logs security is essential for maintaining trust in your logging infrastructure and protecting the valuable insights contained within your log data.