Monitoring on AWS

Cloud Watch Metrics, Cloud Watch Alarms

Learning Outcome

Let’s recall what we already learned

CloudWatch monitors AWS resources
 

It collects metrics and logs
 

Monitoring helps maintain availability and reliability

Understanding the Concept

It shows temperature continuously and rings when temperature becomes too high.

Imagine a digital thermometer with an alarm feature.

Similarly, CloudWatch shows system performance using metrics and triggers alarms when limits are crossed

What are CloudWatch Metrics?

Amazon CloudWatch automatically collects numerical data from AWS resources

Performance            Resource utilization           System behavior

These numerical values are called metrics

Metrics represent:

Examples of Default Metrics

• Network Out

• CPU usage

For EC2 instance

For RDS

Disk Read/Write

CloudWatch automatically

collects these metrics

• CPU Utilization

• Network In

• Database connections

• Free storage space

Types of Metrics

• Automatically provided by AWS
• No additional setup required

• User-defined metrics
• Sent manually to CloudWatch

Default Metrics

 Custom Metrics

Example: Business transactions

Example: EC2 CPU utilization

Important Metric Components

Timestamp

Each metric includes:

Metric name

Value

Why Metrics are Important?

What are CloudWatch Alarms?

CloudWatch alarms monitor metrics and perform actions when thresholds are crossed.

An alarm continuously checks a metric and changes state based on conditions.

Alarm States

1. OK – Metric is within safe limit
    

2. ALARM – Threshold exceeded
    

INSUFFICIENT_DATA – Not enough data

How CloudWatch Alarms Work

Step-by-step:

1. Select a metric (e.g., CPU Utilization)
    

2. Define threshold (e.g., > 80%)
    

3. Set evaluation period
    

4. Define action (send notification, trigger scaling)

Alarm Actions

When alarm triggers, it can:

• Send notification (SNS)
   

• Trigger Auto Scaling
   

• Stop or terminate EC2 instance
   

Recover instance

Why Alarms are Important?

• Provide automatic alerting
   

• Reduce downtime
   

• Enable automation
   

• Improve reliability

Without alarms:

• Teams must manually check dashboards
   

• Problems may go unnoticed

Practical Example

Example Scenario:

• EC2 CPU > 80% for 5 minutes
   

• Alarm triggers
   

• Auto Scaling launches new instance
   

Result:
System remains stable.

Key Points

• Metrics are numerical performance data
   

• Default metrics are automatic
   

• Custom metrics require manual publishing
   

• Alarms work based on thresholds
   

• Alarm states: OK, ALARM, INSUFFICIENT_DATA
   

• Alarm actions can automate responses

Summary

 What are CloudWatch metrics?

 What are CloudWatch metrics?