Blue Team · 9 min read · Log Academy training team
What is SIEM and how does it work? A simple guide with Wazuh and ELK
SIEM (Security Information and Event Management) is a security system that collects logs from a company’s servers, computers and network devices in one place, analyzes them against rules and raises an alert when something suspicious happens. Put simply, a SIEM finds the sequence that looks like an attack among thousands of ordinary events. A SOC analyst spends most of the working day in a SIEM console; Wazuh and ELK are open-source examples.
- A SIEM does four jobs: it collects logs, normalizes them, correlates them with rules and raises alerts.
- The power of a SIEM lies not in a single event but in connecting events from different sources.
- Wazuh and ELK are free and open source, so you don’t need a license to learn SIEM.
- A SIEM does not stop an attack on its own; the analyst makes the call.
- The fastest way to learn SIEM is to send logs in your own lab and hunt for events.
What is SIEM and why do you need it?
A mid-sized company generates hundreds of thousands of log entries every day: user logins, process launches, firewall decisions, web requests. These logs live on different devices in different formats. Opening and reading them one by one is impossible.
A SIEM solves this problem. It gathers all logs in a central place, puts them into a common format and cuts search time down to seconds. More importantly, its rules catch connections a human would miss: for example, several failed logins to the same account late at night, followed by a successful login and the immediate creation of a new user.
How does SIEM work?
- Collection: logs from Linux, Windows, network devices and applications are sent to the SIEM via agents or syslog.
- Parsing and normalization: logs in different formats are split into the same fields — time, user, IP address, event type.
- Enrichment: extra context is attached to the log, such as which server an IP address belongs to.
- Correlation: rules combine several events and flag a suspicious sequence.
- Alerting: when a rule fires, an alert lands in the analyst’s queue with a priority assigned.
- Investigation and reporting: the analyst checks whether the alert is real or a false positive and documents the outcome.
The weakest link in this chain is usually step two. If a log isn’t parsed correctly, the username and IP address don’t end up in separate fields, and no rule will fire. That’s why hands-on SIEM work is far more about data quality than it first appears.
What are the main components of a SIEM?
| Component | What it does | Example in Wazuh / ELK |
|---|---|---|
| Agent / shipper | Reads logs at the source and forwards them to the central system | Wazuh agent, Beats |
| Parser | Splits the log into fields | Wazuh decoders, Logstash |
| Storage and index | Stores logs for fast search | Wazuh indexer, Elasticsearch |
| Rule engine | Finds suspicious sequences | Wazuh rules, Elastic detection rules |
| Dashboard | Visual overview and search interface | Wazuh dashboard, Kibana |
Wazuh vs ELK: what’s the difference?
ELK (Elasticsearch, Logstash, Kibana) is essentially a general-purpose log platform: it offers powerful search and visualization, but security features have to be built on top of it. Wazuh, by contrast, is designed for security: it ships with an agent, ready-made detection rules, file integrity monitoring and vulnerability detection.
| Criterion | Wazuh | ELK |
|---|---|---|
| Main purpose | Security monitoring | General log search and analytics |
| Built-in security rules | Yes, included out of the box | Added via Elastic Security |
| Agent | Wazuh agent | Beats / Elastic Agent |
| For learning | A quick start in SOC work | A deep understanding of log processing |
| Cost | Open source | Basic version is free |
Beginners don’t need to see the two as rivals. Indexing and the dashboard inside Wazuh are also based on OpenSearch, so the search logic you learn in ELK carries over to Wazuh. Job postings also mention commercial SIEMs such as Splunk or QRadar; the concepts are the same, and the main difference is the query language.
Which logs should you send to a SIEM first?
Trying to collect every log at once usually fails: storage fills up, search slows down and the important signal gets lost in the noise. A better approach is to start with the sources where attacks most often leave traces.
- Authentication logs: Windows domain logins, Linux SSH logins, VPN and email logins.
- Process creation: which program was launched on Windows and with which parameters, captured by Sysmon.
- Firewall and proxy logs: which internal machine connects to which external address.
- DNS queries: requests to malicious domains often show up here first.
- Server and application logs: requests to the web server and error messages.
- Account changes: new users, group memberships and permission changes.
What is the difference between SIEM, EDR and SOAR?
These three terms are often confused because all of them are used in a SOC. The difference lies in what they look at and what they do.
| Tool | What it looks at | Main job |
|---|---|---|
| SIEM | Logs from all sources | Collect, correlate, raise alerts |
| EDR | Individual computers and servers | Monitor endpoint behavior and stop processes |
| SOAR | Alerts from the SIEM and other tools | Automate repetitive response steps |
In a real SOC these tools work together. The EDR sends its events to the SIEM, the SIEM combines them with other sources, and the SOAR takes automatic action on a confirmed alert — for example, locking an account or opening a ticket. For a beginner, the SIEM matters most, because the logic of the other two runs through it.
How do you investigate a typical SIEM alert?
- Read what the alert says: which rule, which machine, which user, and when.
- Look at the events before and after the alert for that user and machine.
- Check whether the behavior is normal: does this user usually log in at this hour, from this location?
- Check related indicators: has this IP address, domain or file hash been seen before?
- Decide: false positive, an event that needs further investigation, or an incident that requires escalation.
- Record the decision and the reasoning so the next analyst doesn’t repeat the same work.
How do you write a SIEM detection rule?
A rule is a formal way of writing down a question: “if these events happen in this order within this time window, is it suspicious?” Every SIEM has its own rule format. That’s why defenders often use the Sigma format: you write a rule once in Sigma and then convert it into the language of different SIEMs.
- Tag the rule with a MITRE ATT&CK technique so it’s clear which attack behavior it covers.
- First reproduce the attack in a lab and confirm that the rule actually fires.
- Measure how often the rule fires by mistake during normal activity — false positives eat up the analyst’s time.
- Add a short description and “what to do when this alert fires” steps to the rule.
What are the most common mistakes when working with a SIEM?
- Collecting everything but analyzing nothing: unnecessary logs both fill up storage and slow down search.
- Enabling ready-made rules without testing them: hundreds of meaningless alerts flood the queue and the analyst misses the important signal.
- Ignoring time synchronization: if server clocks differ, the sequence of events looks wrong.
- Not setting up extra telemetry on Windows: the default logs give little detail, while Sysmon records process and network events in depth.
How we do it in our lab
In the Log Academy Blue Team program, SIEM is set up in the third module: in a personal virtual lab, students build log collection, parsing, a dashboard and a report on ELK or Wazuh. They then trace a suspicious authentication chain — several failed logins, followed by a successful login and the creation of a new user — and write a Sigma rule for that sequence. Where students get stuck most is not log collection but correlation: seeing that two ordinary events from two separate sources together point to an attack.
Where should you start learning SIEM?
- Learn networking basics: see what TCP/IP, DNS and HTTP traffic looks like in Wireshark.
- Learn where Linux and Windows logs are stored and which event IDs matter.
- Set up Wazuh or ELK in a lab and send logs from two machines.
- Create a simple scenario and find it in the SIEM with a search.
- Write a rule for that scenario, reduce false positives and prepare a short incident report.
You don’t need to know how to code
You don’t have to be a programmer to get started with SIEM. At the first stage, the key skill is reading a log and asking “is this normal?” The query language and simple scripts come later, with practice.
Frequently asked questions
- How would you explain SIEM in one sentence?
- A SIEM is a security system that collects all of a company’s logs in one place, uses rules to find suspicious events and sends alerts to an analyst.
- What is the difference between SIEM and SOC?
- A SOC is a security operations center, meaning the team and its processes. A SIEM is the main tool that team uses.
- Is Wazuh free?
- Yes. Wazuh is open source and can be installed on your own server or in your lab without a license.
- How long does it take to learn SIEM?
- You can learn the basic principles and simple searches in a few weeks. Writing rules and investigating incidents, however, takes months of lab practice.
- Where can you learn SIEM hands-on?
- In your own home lab or in a hands-on course. In the 6-month Log Academy Blue Team program, SIEM, Sigma rules and incident response are covered in a personal lab, in both in-person and online formats. The first lesson is free.