AWS Knowledge Hub

English

Tiếng Việt

English

Tiếng Việt

Appearance

Amazon CloudWatch ​

Amazon CloudWatch is a monitoring and observability service that provides data and actionable insights for AWS, hybrid, and on-premises applications and infrastructure resources.

Overview ​

CloudWatch collects monitoring and operational data in the form of logs, metrics, and events, and provides you with a unified view of AWS resources, applications, and services.

Key Features ​

  • Metrics: Collect and track metrics
  • Logs: Centralized log management
  • Alarms: Automated actions based on thresholds
  • Dashboards: Customizable visualization
  • Events: Event-driven automation
  • Insights: Log analytics and queries
  • Synthetics: Monitor endpoints and APIs
  • ServiceLens: End-to-end application monitoring

Core Concepts ​

Metrics ​

Time-ordered data points that represent a measurement over time. Examples:

  • CPU utilization
  • Network traffic
  • Request count
  • Error rate

Namespaces ​

Containers for metrics. AWS services use namespaces like AWS/EC2, AWS/S3.

Dimensions ​

Name-value pairs that uniquely identify a metric. Examples:

  • Instance ID
  • Auto Scaling Group
  • Environment

Alarms ​

Monitor metrics and perform actions when thresholds are crossed.

Dashboards ​

Customizable home pages for monitoring resources.

Metrics ​

Standard Metrics ​

  • Automatically collected by AWS services
  • No additional configuration
  • 1-minute or 5-minute granularity
  • Free tier available

Custom Metrics ​

  • Metrics you publish
  • Any application or business metric
  • 1-second to 1-hour granularity
  • Charged per metric

Metric Math ​

  • Perform calculations on metrics
  • Create composite metrics
  • Statistical functions
  • Time-series functions

Anomaly Detection ​

  • Machine learning-based detection
  • Identify unusual patterns
  • Reduce false alarms
  • Automatic baseline learning

Logs ​

Log Groups ​

Containers for log streams. Organize logs by application or service.

Log Streams ​

Sequences of log events from the same source. Instances, containers, etc.

Log Insights ​

  • Query log data using SQL-like syntax
  • Real-time log analysis
  • Visualize query results
  • Save queries for reuse

Metric Filters ​

  • Extract metrics from logs
  • Create CloudWatch metrics from log data
  • No code changes required
  • Cost-effective monitoring

Alarms ​

Alarm States ​

  • OK: Within threshold
  • ALARM: Threshold breached
  • INSUFFICIENT_DATA: Not enough data

Alarm Actions ​

  • SNS Notifications: Send notifications
  • Auto Scaling: Scale resources
  • EC2 Actions: Stop, terminate, reboot instances
  • Lambda: Invoke functions
  • Systems Manager: Run automation

Composite Alarms ​

  • Combine multiple alarms
  • Reduce alarm noise
  • Complex alerting logic
  • AND/OR conditions

Dashboards ​

Custom Dashboards ​

  • Visualize metrics and logs
  • Multiple widgets
  • Auto-refresh
  • Share with team

Widget Types ​

  • Line charts
  • Number widgets
  • Logs table
  • Alarm status

Events (EventBridge) ​

Event Rules ​

  • Match incoming events
  • Route to targets
  • Schedule-based rules
  • Event pattern matching

Targets ​

  • Lambda functions
  • SNS topics
  • SQS queues
  • Step Functions
  • Many more

CloudWatch Insights ​

Container Insights ​

  • Monitor containerized applications
  • ECS, EKS, Fargate support
  • Performance metrics
  • Log aggregation

Lambda Insights ​

  • Monitor Lambda functions
  • Performance metrics
  • Error tracking
  • Cold start analysis

Contributor Insights ​

  • Identify top contributors to metrics
  • Anomaly detection
  • Cost optimization insights

Best Practices ​

  • Set Up Alarms: For critical metrics
  • Use Dashboards: For visibility
  • Enable Logging: For troubleshooting
  • Use Metric Filters: Extract metrics from logs
  • Implement Log Retention: Control costs
  • Use Anomaly Detection: Reduce false alarms
  • Monitor Costs: CloudWatch can be expensive
  • Use Log Insights: For log analysis
  • Set Up Composite Alarms: Reduce noise
  • Regular Reviews: Review and optimize alarms

Use Cases ​

  • Application Monitoring: Monitor application performance
  • Infrastructure Monitoring: Monitor AWS resources
  • Log Management: Centralized log storage and analysis
  • Alerting: Automated notifications
  • Capacity Planning: Track resource utilization
  • Troubleshooting: Debug issues with logs and metrics

Cost Optimization ​

  • Use standard metrics (free tier)
  • Set appropriate log retention
  • Use metric filters instead of custom metrics when possible
  • Delete unused dashboards
  • Use appropriate metric granularity
  • Monitor CloudWatch costs
  • Use Log Insights efficiently

Limits ​

  • Metrics per Namespace: Unlimited
  • Alarms per Account: 5,000
  • Dashboards per Account: 3,000
  • Log Retention: 1 day to forever
  • Metric Data Retention: 15 months