CloudConsole

CloudWatch alarm for Amazon Cognito FederationThrottles

Updated 16 November 2026 · How we write and check guides

Short answer

FederationThrottles counts federated sign-in requests, through SAML, OIDC or social identity providers, that a Cognito user pool throttled. AWS recommends an alarm on its Sum per UserPool, UserPoolClient and IdentityProvider, above a threshold set from your history, for 5 of 5 one-minute periods.

Chart of the FederationThrottles alarm: Sum per 1-minute period, ALARM when 5 of 5 datapoints are > 10
How the recommended alarm evaluates, with AWS's settings from this page and the page's example threshold. The metric values are illustrative.

How federation throttling differs

When users sign in through an external identity provider, the user pool's managed login pages handle the redirect, receive the provider's assertion or code, and issue Cognito tokens. That flow has its own quota category, separate from direct username-and-password sign-in, and its throttles are counted in FederationThrottles.

The extra IdentityProvider dimension is the provider name configured in the pool, for example Google or CorpSAML. That lets you alarm on the provider your workforce depends on without noise from others.

Typical triggers:

  • a workforce app where everyone signs in through corporate SSO at the start of the day;
  • a redirect loop in a single-page app that restarts the federation flow on every page load;
  • a session timeout change forcing everyone to re-authenticate at once.

Recommended alarm settings

SettingValue
NamespaceAWS/Cognito
MetricFederationThrottles
DimensionsUserPool, UserPoolClient, IdentityProvider
StatisticSum
Period60 seconds
Datapoints to alarm5
Evaluation periods5
Comparison operatorGreaterThanThreshold
ThresholdDepends on your situation: above your normal level of throttled federation requests

As given on AWS's CloudWatch recommended alarms page.

Choosing the threshold

Federation has naturally spiky traffic, so a zero threshold tends to alarm at every morning peak. Pull a month of one-minute sums for FederationThrottles and FederationSuccesses per provider. If throttles cluster at a predictable daily peak and users get through on retry, set the threshold above that peak. If throttles appear outside peaks, or the ratio of throttles to successes is growing week by week, treat it as under-provisioning and keep the threshold low. Leave missing data as notBreaching.

Create it

# 10 per minute is an example; set it from your own history
aws cloudwatch put-metric-alarm \
  --alarm-name userpool-corpsaml-federation-throttles \
  --namespace AWS/Cognito \
  --metric-name FederationThrottles \
  --dimensions Name=UserPool,Value=eu-west-2_AbCdEf123 Name=UserPoolClient,Value=1h2j3k4l5m6n7o8p9q0r1s2t3u Name=IdentityProvider,Value=CorpSAML \
  --statistic Sum \
  --period 60 \
  --evaluation-periods 5 \
  --datapoints-to-alarm 5 \
  --threshold 10 \
  --comparison-operator GreaterThanThreshold \
  --treat-missing-data notBreaching \
  --alarm-actions arn:aws:sns:eu-west-2:123456789012:ops-alerts \
  --region eu-west-2

List provider names with aws cognito-idp list-identity-providers --user-pool-id eu-west-2_AbCdEf123.

When it fires

  1. Look for a loop. If one browser or IP appears over and over in your app's logs, a front-end bug is restarting the flow. Fixing it usually ends the throttling.
  2. Check recent session changes, such as shorter refresh token validity, which push more users back through the identity provider.
  3. Stagger planned re-authentication, for example after rotating signing certificates, instead of invalidating every session at once.
  4. Check the quota for the federation category in Quotas in Amazon Cognito and request an increase if real demand is higher.
  5. Tell users to retry rather than repeatedly clicking sign-in; each click starts a new flow.

Questions

What does the IdentityProvider dimension contain?

The name of the identity provider as configured in the user pool, such as Google or the name you gave a SAML provider.

Are federation and normal sign-in throttled separately?

Yes. Federated sign-ins have their own quota category and their own throttle metric, FederationThrottles.

Why do throttles spike at 9am?

Workforce apps see most users authenticate through SSO in a short window. If retries succeed, set the threshold above that daily peak.

When this happens away from your desk

When SSO sign-in fails, the whole company notices within minutes. CloudConsole, a phone app for AWS, is built to show this alarm per identity provider on a phone so you can tell whether it is a quota or a loop.

Did it start after a deploy? What changed? on the app’s alarm screen lists the deploys and configuration changes from the hour before the alarm fired.

About the app → · Download on the App Store

Sources