PagerDuty and AWS together turn raw operational data into actionable information by automating the triage, routing, and escalation of alerts generated by AWS services. You connect AWS CloudWatch, CloudTrail, and other monitoring tools to PagerDuty, which then deduplicates, correlates, and notifies the right on-call engineer—transforming a stream of metrics and logs into a clear, prioritized incident response.
Formula
The calculator above estimates how many alerts in a given time window are truly actionable after removing duplicates, and how many require escalation. The formulas are:
- Actionable alerts = Alerts received × (1 − Duplicate percentage / 100)
- Escalated alerts = Alerts received × (Escalation percentage / 100)
- Noise (duplicates) = Alerts received × (Duplicate percentage / 100)
Variables: a = total alerts received, d = percentage of duplicate alerts, e = percentage of alerts that require escalation. All inputs are numbers you provide; the output is a count.
Worked Example
Using the default values: 100 alerts received in 24 hours, 40% duplicates, 10% escalated.
- Actionable = 100 × (1 − 40/100) = 100 × 0.6 = 60
- Escalated = 100 × (10/100) = 10
- Noise = 100 × (40/100) = 40
So out of 100 alerts, 60 are unique and actionable, 10 need escalation, and 40 are duplicates that can be suppressed.
Why This Matters: The Operational Data Problem
Modern cloud infrastructure generates a massive volume of operational data—metrics from CloudWatch, logs from CloudTrail, and events from services like EC2, Lambda, and RDS. Without a system to interpret this data, teams face alert fatigue, missed critical incidents, and slow response times. PagerDuty acts as the intelligence layer that filters, correlates, and routes this data to the right person at the right time, turning raw signals into a clear, prioritized action list.
How PagerDuty and AWS Integrate
The integration works through a series of steps:
- Data collection: AWS services emit metrics and events to CloudWatch and CloudTrail.
- Alert generation: CloudWatch Alarms trigger on thresholds (e.g., CPU > 80% for 5 minutes).
- Event forwarding: An Amazon Simple Notification Service (SNS) topic sends the alarm notification to a PagerDuty endpoint via a webhook or the AWS Lambda integration.
- Processing in PagerDuty: PagerDuty receives the event, deduplicates it (using a deduplication key), correlates it with related incidents, and applies routing rules based on the service and severity.
- Notification and escalation: The on-call engineer is notified via push, SMS, or phone call. If they don’t acknowledge within a set time, the incident escalates to the next level.
This pipeline ensures that an alarm like “RDS CPU Utilization > 90%” becomes a PagerDuty incident with the exact database instance, the severity, and a link to the CloudWatch graph—not just a raw number in a dashboard.
Key Features That Make It Actionable
Several PagerDuty features are critical to converting data into action:
- Deduplication: Repeated alerts from the same AWS alarm are grouped into one incident, reducing noise.
- Event correlation: Related alerts (e.g., multiple EC2 instances failing) are merged into a single incident, providing context.
- Service routing: Alerts are routed to the correct team based on the AWS service and custom rules.
- Severity mapping: AWS alarm severity (e.g., OK, ALARM) is mapped to PagerDuty urgency (low, high) to prioritize response.
- Runbook automation: PagerDuty can trigger AWS Systems Manager Automation documents or Lambda functions to take immediate action, such as restarting an instance or scaling up.
These features turn a raw CloudWatch alarm into a structured incident with a clear owner, priority, and next step.
Common Problems and Misconceptions
One common misconception is that PagerDuty is just an alerting tool. In reality, it’s an incident management platform that uses operational data to drive workflows. Another issue is alert fatigue: teams often connect every CloudWatch alarm to PagerDuty, overwhelming on-call staff. The right approach is to filter alerts at the source (e.g., only alarm on meaningful thresholds) and use PagerDuty’s deduplication and suppression rules.
Another problem is misconfigured SNS topics or missing IAM permissions, which can cause alerts to never reach PagerDuty. Always test the integration after setup. Also, remember that PagerDuty is not a monitoring tool—it doesn’t collect data itself; it depends on AWS services to send it data.
How to Choose What to Send to PagerDuty
Not all operational data should become incidents. A good rule is to send only events that require human intervention or immediate action. For example:
- Send: CloudWatch alarms for high error rates, instance failures, or security threats.
- Don’t send: Routine metrics like average CPU usage or minor log messages that don’t indicate a problem.
Use CloudWatch composite alarms to combine multiple conditions, reducing false positives. Also, set appropriate thresholds and evaluation periods to avoid flapping.
Costs and Alternatives
PagerDuty pricing is based on the number of users (on-call responders) and features, with a free tier for limited use. AWS itself offers Amazon CloudWatch Alarms and AWS Incident Manager (part of AWS Systems Manager) as a native alternative. AWS Incident Manager provides similar incident response capabilities but is less mature than PagerDuty in terms of integrations and scheduling. Other alternatives include Opsgenie and VictorOps. The choice depends on your team’s size, existing tooling, and budget.
Frequently Asked Questions
Can PagerDuty automatically fix AWS issues?
Yes, through integrations with AWS Lambda and Systems Manager. You can create PagerDuty rules that trigger a Lambda function to perform a predefined action, such as restarting an EC2 instance or increasing a DynamoDB table’s capacity. This is often called “runbook automation” and can resolve common issues without human intervention.
What is the difference between CloudWatch Alarms and PagerDuty incidents?
CloudWatch Alarms are raw monitoring signals—they tell you a metric crossed a threshold. PagerDuty incidents are managed workflows that include alert deduplication, on-call routing, escalation policies, and post-incident analysis. CloudWatch is the data source; PagerDuty is the action layer.
Do I need to use AWS Lambda for the integration?
No, but it’s common. You can connect CloudWatch Alarms directly to PagerDuty via an SNS topic with a PagerDuty webhook endpoint. Lambda is useful when you need to transform the event data before sending it to PagerDuty, such as adding custom fields or filtering alerts.
How do I reduce alert noise from AWS?
Start by adjusting CloudWatch alarm thresholds and evaluation periods to reduce false positives. In PagerDuty, use deduplication keys to group repeated alerts, create suppression rules for known maintenance windows, and set up event rules to drop non-critical events. Also, regularly review alarm metrics to ensure they still make sense.
Is PagerDuty the only option for AWS incident management?
No. AWS offers its own Incident Manager, which integrates with CloudWatch and other AWS services. Other third-party tools like Opsgenie and VictorOps also provide similar functionality. PagerDuty is popular for its extensive integration ecosystem and advanced features, but the best choice depends on your specific needs and budget.
Related
Formula used
in t | Time window (hours) | hours | 24 | 1 | 720 in a | Alerts received | count | 100 | 1 | 100000 in d | Duplicate alerts (%) | % | 40 | 0 | 100 in e | Escalated alerts (%) | % | 10 | 0 | 100 out actionable | Actionable alerts | count | a * (1 - d/100) | 0 out escalated | Escalated alerts | count | a * (e/100) | 0 out noise | Noise (duplicates) | count | a * (d/100) | 0
Each line is one input (in, sel, list) or one result (out: label, unit, formula, decimals). How the calculators work
Method: the standard formula for this calculation, rounded as each result states. See how the calculators work or report an error.
More cards in this drawer
- How to Choose an Amplifier for Magico A3 Speakers
- How to Ban a Player from Your FiveM Server
- Pokemon Go Friend Codes: How to Add & Use Them33.33 XP/day
- Google Algorithm Change: Original Reporting Boost
- Create Folder for Selected Item on Mac: Full Guide
- Data Center Location Questions for AI Workloads