Understanding CloudWatch Math Metrics

September 13, 2023

Kalpana Venkatesan

Creating and working with AWS CloudWatch math metrics can be a powerful way to gain insights from your AWS resources. Math metrics allow you to perform calculations on existing metrics to derive new information or trigger alarms based on specific conditions. In this blog post, we’ll explore the concept of CloudWatch math metrics and provide a sample scenario to demonstrate their usage.

Understanding CloudWatch Math Metrics

AWS CloudWatch provides a feature called “Math Expressions” that enables you to create custom math metrics based on existing metrics. You can use math expressions to perform calculations, aggregations, and comparisons across multiple metrics. This is particularly useful for creating composite metrics or triggering alarms based on calculated values.

Sample Scenario: Calculating Transit gateway Packet loss Percentage

Let’s consider a common use case where you want to calculate the error rate of a transit gateway packet loss by dividing the number of packets lost by the total number of requests over a specific time period. Here’s how you can achieve this using CloudWatch math metrics:

Prerequisites:
- You should have already set up CloudWatch to monitor your application and collect metrics.
- Ensure that you have below CloudWatch metrics available:
- PacketDropCountNoRoute (Count of Packets lost without a route)
- PacketsIn (Count of total packets received)
- PacketsOut (Count of total packets sent)

Create a Math Expression:
- Go to the AWS CloudWatch console.
- In the navigation pane, select “Metrics.”
- Under “All metrics,” select the namespace where your metrics reside.
- Choose “Math expressions” from the dropdown menu.
- Click “Create math expression.”
- Define your math expression as FLOOR( (PacketDropCountNoRoute *100) / PacketsIn+ PacketsOut)) to calculate the error rate as a percentage.

Create an Alarm:
- After defining the math expression, you can create an alarm based on the calculated error rate.
- Set thresholds and actions for the alarm. For example, you can trigger an SNS notification or an AWS Lambda function if the error rate exceeds a certain threshold.

Visualisation:
- You can also visualise the math expression as a custom graph on your CloudWatch dashboard to monitor the error rate over time.

By following these steps, you’ve created a CloudWatch math metric that calculates the error rate of your application based on the existing metrics for errors and requests. This enables you to proactively monitor and respond to changes in your application’s performance.

Math Metric in Yaml:

TGWPacketDropMetrics:
Type: AWS::CloudWatch::Alarm
Condition: TGWEnabled
Properties:
Metrics:
– Expression: “FLOOR((pktDrpCnt*100)/(pktIn+pktOut))”
Id: “ppercent”
Label: “PercentOfPktsDrpPerPeriod”
ReturnData: true
– Id: “pktDrpCnt”
Label: “TotalPacketsDropped”
MetricStat:
Metric:
Dimensions:
– Name: TransitGateway
Value: !Ref TGWID
MetricName: PacketDropCountNoRoute
Namespace: AWS/TransitGateway
Period: 60
Stat: Sum
Unit: Count
ReturnData: false
– Id: “pktIn”
Label: “TotalPacketsIn”
MetricStat:
Metric:
Dimensions:
– Name: TransitGateway
Value: !Ref TGWID
MetricName: PacketsIn
Namespace: AWS/TransitGateway
Period: 60
Stat: Sum
Unit: Count
ReturnData: false
– Id: “pktOut”
Label: “TotalPacketsOut”
MetricStat:
Metric:
Dimensions:
– Name: TransitGateway
Value: !Ref TGWID
MetricName: PacketsOut
Namespace: AWS/TransitGateway
Period: 60
Stat: Sum
Unit: Count
ReturnData: false
EvaluationPeriods: 3
Threshold: !Ref TGWPacketDropThreshold
ComparisonOperator: GreaterThanThreshold
AlarmActions:
– !GetAtt SNOWAlertRelay.Outputs.SNSArn

Conclusion

In conclusion, AWS CloudWatch math metrics provide a powerful way to derive valuable insights from your existing metrics and take automated actions based on custom calculations.

Enjoyed this blog?

Share it with your network!