Amazon CloudWatch Metrics

1️⃣ What are Amazon CloudWatch Metrics?

  • Amazon CloudWatch Metrics are time-ordered sets of data points that represent the performance of AWS resources and applications.
  • Each metric has:
    • Namespace (e.g., AWS/EC2)
    • Metric Name (e.g., CPUUtilization)
    • Dimensions (e.g., InstanceId=i-12345)
    • Timestamps
    • Unit of measure (e.g., percent, bytes, seconds)

2️⃣ Key Features

  • Metrics are collected automatically for many AWS services.
  • Metrics are stored for 15 months for historical analysis.
  • You can publish custom metrics from your applications or on-premises systems.
  • Supports standard resolution (1-minute) and high-resolution (1-second) metrics.

3️⃣ Common AWS Service Metrics

Service Metric Examples
EC2 CPUUtilization, NetworkIn, DiskReadOps
S3 BucketSizeBytes, NumberOfObjects
RDS DatabaseConnections, ReadIOPS
Lambda Invocations, Duration, Errors
DynamoDB ReadThrottleEvents, WriteCapacityUtilization

4️⃣ CloudWatch Custom Metrics

You can create your own metrics using:

aws cloudwatch put-metric-data \
  --namespace "MyAppNamespace" \
  --metric-name "PageLoadTime" \
  --value 2.5 \
  --unit Seconds

Example use case: Track application latency or business KPIs like "OrdersPlaced".

5️⃣ Metric Granularity & Retention

Resolution Type Data Storage Duration
1-second (High-Res) 3 hours
1-minute (Standard) 15 days
5-minute average 63 days
1-hour average 15 months

6️⃣ How Metrics Work in Monitoring

  1. AWS services send metrics to CloudWatch.
  2. CloudWatch stores them in a namespace.
  3. You can:
    1. Create alarms to trigger notifications or actions (e.g., Auto Scaling).
    2. Visualize metrics using CloudWatch Dashboards.
    3. Set anomaly detection to identify unusual patterns.

✅ In Short:

  • CloudWatch Metrics = Key performance indicators for AWS resources and custom apps, stored over time, used for monitoring, dashboards, and automated scaling actions.

✅ AWS CloudWatch Metrics Cheatsheet

CloudWatch Metrics are time-series data points that help monitor AWS resources and applications, visualize performance, set alarms, and automate scaling.

1️⃣ EC2 (Amazon Elastic Compute Cloud)

Metric Name Unit Description
CPUUtilization Percent (%) CPU usage across all cores
NetworkIn Bytes Amount of incoming traffic
NetworkOut Bytes Amount of outgoing traffic
DiskReadOps Count Number of read operations on instance store
DiskWriteOps Count Number of write operations on instance store
StatusCheckFailed Count (0/1) Instance or system reachability check failures

2️⃣ Lambda (AWS Lambda Functions)

Metric Name Unit Description
Invocations Count Number of times the function is executed
Duration Milliseconds Execution time of the function
Errors Count Number of failed function executions
Throttles Count Number of invocation requests throttled
IteratorAge Milliseconds Age of the last record processed for stream-based invocations
ConcurrentExecutions Count Number of functions running at the same time

3️⃣ S3 (Amazon Simple Storage Service)

Metric Name Unit Description
BucketSizeBytes Bytes Total storage size of objects in a bucket
NumberOfObjects Count Number of stored objects
AllRequests Count Number of requests to S3 bucket
4xxErrors Count Number of client errors
5xxErrors Count Number of server errors
FirstByteLatency Milliseconds Time taken from request to first byte returned

4️⃣ RDS (Amazon Relational Database Service)

Metric Name Unit Description
CPUUtilization Percent (%) CPU usage of DB instance
DatabaseConnections Count Number of active DB connections
ReadIOPS Count/sec Read operations per second
WriteIOPS Count/sec Write operations per second
FreeStorageSpace Bytes Remaining available storage
ReplicaLag Seconds Replication delay in a read replica

5️⃣ DynamoDB (Amazon DynamoDB Table)

Metric Name Unit Description
ConsumedReadCapacityUnits Count/sec Read capacity used
ConsumedWriteCapacityUnits Count/sec Write capacity used
ReadThrottleEvents Count Number of read requests throttled
WriteThrottleEvents Count Number of write requests throttled
SuccessfulRequestLatency Milliseconds Time taken to complete requests
SystemErrors Count Internal server errors from DynamoDB

📌 Key Notes:

  • Metrics are stored for 15 months for historical analysis.
  • Can set alarms to trigger Auto Scaling or SNS notifications.
  • Supports high-resolution metrics (1-second intervals) for faster reaction time.
  • Custom metrics can be created for application-specific KPIs using:
aws cloudwatch put-metric-data --namespace "MyApp" --metric-name "PageLoadTime" --value 2.5 --unit Seconds

🚀 Pro Tip for Certification:

  • EC2: CPUUtilization > 80% ⇒ Trigger Auto Scaling.
  • Lambda: Monitor Throttles for concurrency limits.
  • S3: Use 4xxErrors + 5xxErrors to troubleshoot access issues.
  • RDS: Watch ReplicaLag for replication delays.
  • DynamoDB: Keep an eye on throttled events for scaling needs.


Back to blog

Leave a comment