API activity audit logs
What's different
Records who did what, and when, in the control plane.
Security & Identity · API activity audit logs
API activity audit logs, side by side on AWS, Azure and Google Cloud. They're a near 1:1 match, so a migration is mostly configuration.
API activity audit logs
Records who did what, and when, in the control plane.
| Aspect | AWS | Azure | Google Cloud |
|---|---|---|---|
| On by default | Event history: last 90 days of management events | Activity log: 90 days of subscription-level events | Admin Activity logs: always on, kept 400 days, can't be disabled |
| Data-plane events (reads/writes) | Data events: opt-in, extra cost | Resource logs: enable with diagnostic settings | Data Access logs: off by default (except BigQuery), extra cost |
| Keeping logs longer | Create a trail to S3, or use CloudTrail Lake | Diagnostic settings to Log Analytics, Storage or Event Hubs | Log sinks to Cloud Storage, BigQuery or Pub/Sub |
| Organisation-wide | Organization trail across all accounts | Deploy diagnostic settings to every subscription (e.g. with Azure Policy) | Aggregated sinks at organisation or folder level |
You're on AWS: create one organization trail to S3 on day one, since event history only covers 90 days.
You're on Azure: export the activity log with diagnostic settings, or you lose it after 90 days.
You're on Google Cloud: Admin Activity is covered for 400 days by default; turn on Data Access logs only where you need them (they cost extra).
Details for AWS CloudTrail verified November 2026 against official documentation: AWS CloudTrail · Azure Monitor · Google Cloud: Cloud Audit Logs best practices.