What is a SIEM and how does it detect attacks?
A SIEM (Security Information and Event Management) is a central system that collects logs from laptops, servers, firewalls and cloud apps, and runs rules on them to raise alerts about possible attacks.

SIEM
Every failed Windows sign-in writes a record called event 4625. It holds the username, the time and the source address. One record on its own means little. So companies ship logs from laptops, servers, firewalls and cloud apps to one central system. That system is a SIEM (Security Information and Event Management). NIST (the National Institute of Standards and Technology) describes it as providing centralized logging capabilities.
A SIEM finds and reports. It doesn’t fight back by itself. When a company wants a fast reaction, it adds automation, often called SOAR (Security Orchestration, Automation and Response). Automation rules and playbooks can act on an alert, for example by blocking an address on the firewall. Without that, a person has to read the alert and respond.
What a SIEM actually does
First it collects. Agents on machines, APIs from cloud apps and the syslog protocol carry the logs in. Then it normalizes them. Every vendor writes logs in its own format, so the SIEM rewrites them into one structure. Now a firewall entry and a Windows entry line up side by side.
On Windows, event 4625 means ‘An account failed to log on’. It records fields such as TargetUserName, IpAddress, LogonType and SubStatus. A SubStatus of 0xC000006A means a bad password, and LogonType 3 means a network logon. Event 4624 means an account logged on successfully. Detection rules search these events on a schedule. In Microsoft Sentinel, for example, scheduled rules can run from every 5 minutes to every 14 days.
How a SIEM detects password guessing
- Count the failed logons. A rule counts
4625events. MITRE’s example alerts after about 10 to 15 failures within 5 to 10 minutes. - Group by source address. The failures are tied to the same source address. So they read as one pattern instead of scattered records.
- Look for the success. Correlation links the clues: many failures, then one
4624success from the same address. - Turn matches into incidents. A match becomes an alert. Related alerts are grouped into one incident for an analyst.
The result is one incident instead of a pile of loose records. The analyst gets one place to start the investigation.
A SIEM is not a blocking tool
A SIEM blocks nothing by itself. It detects, raises the alarm and keeps the logs for investigation. Blocking happens when a person, or an automation playbook linked to the firewall or endpoint tool, acts on the alert. Target’s 2013 breach shows the gap. Its security tool reportedly sent alerts before card data left the network, and no one at headquarters responded. Alerts only help when someone acts on them.
What to do
- Try a mini version at home: open google.com/devices, sign out any device you don’t recognize, then change your password.
- In your Google Account, the same list sits under Security & sign-in, then Your devices, then Manage all devices.
- If you run a SIEM, make sure every alert has someone who responds to it.
- Treat MITRE’s 10 to 15 failures in 5 to 10 minutes as an example to tune, not a standard.
Watch it in 60 seconds
