Correlation Rules and Use Cases
What you'll learn: How to build detection logic in SIEM systems to identify cloud-specific attack patterns like privilege escalation, data exfiltration, unauthorized access, and lateral movement.
What Are Correlation Rules?
Correlation rules are detection patterns that connect multiple log events across time and sources to identify suspicious behavior. Unlike single-event alerts (like "login failed"), correlation rules look for sequences or combinations of events that indicate an active threat.
Think of it like a smoke detector versus a fire investigator. A single alarm might trigger from burnt toast, but an investigator correlating "smoke alarm + broken window + missing valuables" reveals a burglary with arson.
Cloud-Specific Use Cases
Privilege Escalation Detection
Pattern: User assumes role → Creates new admin policy → Attaches policy to their account
- Correlates CloudTrail events:
AssumeRole+CreatePolicy+AttachUserPolicywithin minutes - Flags when elevated permissions are self-granted
Data Exfiltration Detection
Pattern: Large S3 data transfer to unusual destination + API calls from unfamiliar IP
- Correlates S3 access logs showing massive
GetObjectrequests + CloudTrail showingCreateBucketin different region - Triggers when data volume exceeds baseline by 500%
Unauthorized Access Detection
Pattern: Failed login attempts → Successful login from new geography → API calls to sensitive resources
- Correlates authentication logs + VPC Flow Logs + CloudTrail resource access
- Flags "impossible travel" scenarios (Tokyo to London in 10 minutes)
Lateral Movement Detection
Pattern: Compromised EC2 instance → SSH connections to multiple internal IPs → Credential access attempts
- Correlates VPC Flow Logs + application logs + system authentication logs
- Identifies scanning and pivoting behavior across subnets
Key Takeaway: Correlation rules transform isolated log entries into meaningful attack narratives by connecting related events across time, users, resources, and data sources—essential for catching sophisticated cloud threats that span multiple services.