alexsusanu@docs:CloudWatch vs CloudTrail - AWS Services Comparison $
alexsusanu@docs
:~$ cat CloudWatch vs CloudTrail - AWS Services Comparison.md

Home → NOTES → CloudWatch vs CloudTrail - AWS Services Comparison

CloudWatch vs CloudTrail - AWS Services Comparison

category: Security
tags: aws, monitoring, logging, pentesting, cloud-security

General Overview & Core Differences

CloudWatch and CloudTrail are two completely separate AWS services that serve different purposes but can integrate together.

CloudWatch = Performance & Monitoring
- Real-time monitoring and observability
- "How is my infrastructure performing?"
- Metrics, logs, dashboards, and alerts
- Operational health and performance data

CloudTrail = Audit & Security Logging
- API call logging and auditing
- "Who did what, when, and from where?"
- Compliance and security forensics
- User activity and resource change tracking

Key Philosophical Difference

  • CloudWatch: Reactive monitoring - "Something is wrong, fix it"
  • CloudTrail: Proactive auditing - "Someone did something, track it"

Detailed Service Breakdown

CloudWatch Capabilities

  • Metrics Collection
  • CPU, memory, disk, network utilization
  • Custom application metrics
  • Real-time and historical data

  • Automatic scaling triggers

  • Log Management

  • Application logs aggregation
  • System logs from EC2, Lambda, etc.
  • Log insights and searching

  • Log retention policies

  • Alerting & Notifications

  • Threshold-based alarms
  • SNS integration for notifications
  • Auto-scaling triggers
  • Dashboard visualizations

CloudTrail Capabilities

  • API Call Logging
  • Every AWS API call recorded
  • User identity and source IP
  • Request parameters and responses

  • Timestamp and service information

  • Audit Trail Features

  • Management events (control plane)
  • Data events (S3 object access, Lambda execution)
  • Insight events (unusual activity patterns)

  • Multi-region trail support

  • Compliance & Forensics

  • Immutable log records
  • Log file integrity validation
  • Integration with AWS Config
  • Long-term retention options

Service Architecture Differences

CloudWatch Data Flow

AWS Resources → CloudWatch Agent → CloudWatch Service → Dashboards/Alarms
Application Logs → CloudWatch Logs → Log Groups → Insights/Filters

CloudTrail Data Flow

AWS API Calls → CloudTrail Service → S3 Bucket → Analysis Tools
Management Console → API Gateway → CloudTrail Logs → Compliance Reports

Storage & Retention

  • CloudWatch:
  • Metrics: 1 second to 15 months retention
  • Logs: Configurable retention (1 day to indefinite)

  • Real-time streaming available

  • CloudTrail:

  • Events: 90 days in CloudTrail console (free)
  • S3 storage: Indefinite retention (configurable)
  • Log file delivery within 15 minutes

Integration Points

How They Work Together

  • CloudTrail → CloudWatch Logs: Send audit logs to CloudWatch for alerting
  • CloudWatch Alarms: Trigger on specific CloudTrail events
  • Cross-service visibility: Monitor who changed monitoring configurations

Example Integration

{
  "AlarmName": "RootAccountUsage",
  "MetricName": "RootAccountUsageCount",
  "Namespace": "CloudTrailMetrics",
  "Statistic": "Sum",
  "Threshold": 1,
  "ComparisonOperator": "GreaterThanOrEqualToThreshold"
}

Pentesting Applications

CloudWatch for Offensive Security

Information Gathering

  • Target Reconnaissance
  • Identify running services via metrics
  • Spot high-value targets (high CPU/memory usage)
  • Find applications with frequent errors

  • Map resource relationships through dashboards

  • Persistence Indicators

  • Monitor if your access is detected
  • Watch for investigation activities
  • Identify backup and monitoring schedules

Common Misconfigurations

# Check for public CloudWatch dashboards
aws cloudwatch list-dashboards
aws cloudwatch get-dashboard --dashboard-name <name>

# Look for overly permissive log groups
aws logs describe-log-groups
aws logs describe-metric-filters

CloudTrail for Offensive Security

OPSEC & Evasion

  • Audit Trail Analysis
  • Understand what gets logged
  • Identify logging gaps and blind spots
  • Time attacks during high-activity periods

  • Use legitimate-looking API calls

  • Privilege Escalation Intelligence

  • Map user permissions through logged API calls
  • Identify service roles with interesting permissions
  • Find misconfigured cross-account access

Attack Techniques

# Check if CloudTrail is enabled
aws cloudtrail describe-trails
aws cloudtrail get-trail-status --name <trail-name>

# Look for logging misconfigurations
aws cloudtrail lookup-events --lookup-attributes AttributeKey=EventName,AttributeValue=StopLogging

# Find data events configuration (often disabled)
aws cloudtrail get-event-selectors --trail-name <trail-name>

Common Pentesting Scenarios

Scenario 1: CloudTrail Disabled

  • Opportunity: Complete blind spot for defenders
  • Action: Aggressive enumeration and lateral movement
  • Risk: None from audit perspective

Scenario 2: CloudWatch Without Alerting

  • Opportunity: Systems monitored but no one watching
  • Action: Resource abuse, cryptocurrency mining
  • Risk: Eventually discovered through manual review

Scenario 3: Misconfigured Log Storage

# Check S3 bucket permissions for CloudTrail logs
aws s3api get-bucket-acl --bucket <cloudtrail-bucket>
aws s3api get-bucket-policy --bucket <cloudtrail-bucket>

# Look for world-readable CloudTrail logs
aws s3 ls s3://<bucket>/AWSLogs/ --recursive

Key Concepts Summary

  • CloudWatch - Performance monitoring, real-time metrics, operational health
  • CloudTrail - Security auditing, API logging, compliance tracking
  • Separate Services - Different purposes, can work independently
  • Integration Possible - CloudTrail events can trigger CloudWatch alarms
  • Pentesting Value - CloudWatch for recon, CloudTrail for OPSEC

Common Misconfigurations for Exploitation

CloudWatch Misconfigurations

  1. Public Dashboards - Sensitive infrastructure exposed
  2. Overprivileged Metrics - Cross-account metric sharing
  3. Log Group Permissions - World-readable application logs
  4. No Alerting - Monitoring without action

CloudTrail Misconfigurations

  1. Disabled Logging - Complete audit blind spot
  2. Single Region - Multi-region attacks undetected
  3. No Data Events - S3/Lambda activity unlogged
  4. Public S3 Buckets - Historical logs exposed

Best Practices for Defense

  1. Enable Both Services - Complementary coverage
  2. Multi-Region CloudTrail - Global activity visibility
  3. Data Events Logging - S3 and Lambda activity
  4. CloudWatch Alarms - Automated threat detection
  5. Log Integrity - CloudTrail log file validation
  6. Least Privilege - Restrict access to logging services

References / Further Reading

  • AWS CloudWatch Documentation
  • AWS CloudTrail User Guide
  • AWS Well-Architected Security Pillar
  • NIST Cybersecurity Framework - Logging Guidelines
  • SANS Cloud Security Controls
Last updated: 2025-08-26 20:00 UTC