# Correlation
[![[Pasted image 20251104230335.png]]](https://soc-expert.de/From+Logging+to+Alerting+and+Beyond)
**Correlation** refers to the process of **analyzing and linking multiple log events across different systems and sources** to identify patterns that may indicate a security incident or suspicious activity.
Correlation is about **connecting the dots** between seemingly unrelated events to uncover meaningful security insights. A single log entry might not be suspicious on its own, but when combined with other events, it could reveal a threat.
> [!note] Why Is Correlation Important?
>
> - **Reduces noise** by filtering out benign events and focusing on meaningful patterns.
> - **Detects complex attacks** that span multiple systems or timeframes.
> - **Improves incident detection** by identifying behaviors that individual systems might miss.
> - **Supports forensic investigations** by showing how events are related.
## Types of Correlation in SIEM
| Type | Description | Example |
| ---------------------------- | --------------------------------------- | ------------------------------------------------- |
| **Temporal Correlation** | Links events based on time proximity | 5 failed logins within 2 minutes |
| **Spatial Correlation** | Links events based on shared attributes | Same IP address used in multiple events |
| **Cross-source Correlation** | Combines logs from different systems | VPN login + file access + data exfiltration |
| **Behavioral Correlation** | Detects deviations from normal behavior | User downloads 10x more data than usual |
| **Rule-based Correlation** | Uses predefined logic or rules | If A and B happen within X minutes, trigger alert |
| | | |
### Example Scenario
**Without Correlation**:
- A firewall logs a connection from an external IP.
- A server logs a failed login attempt.
- An endpoint logs a file access.
**With Correlation**:
- SIEM links these events and identifies a potential **brute-force attack followed by lateral movement**.
## How SIEMs Use Correlation
- **Correlation Rules**: Define the logic for linking events.
- **Correlation Engines**: Continuously process incoming data to apply rules.
- **Alerting**: When a rule is matched, an alert is generated for investigation.