Explore our new Cybersecurity Terms Glossary, explained by our experts.

View the Glossary View the Glossary
PRIVIA

Cybersecurity Terms Glossary

What Is Zeek?

Zeek watches network traffic and writes structured logs describing every connection, request, and handshake, giving analysts a record of what happened rather than a list of alerts.

Illustration for the cybersecurity term Zeek
Home / Blog / Cybersecurity Terms / What Is Zeek?
17 November 2026

What Is Zeek?

What Is Zeek?

Zeek does something different from the engines it is usually mentioned alongside. It does not primarily look for known-bad patterns. It observes traffic, understands the protocols, and writes a structured record of every connection, DNS query, HTTP request, TLS handshake, SSH session, SMB operation, and file transfer it sees. The output is a description of what the network did, available whether or not anything was flagged.

Vern Paxson created it at Lawrence Berkeley National Laboratory in 1995 under the name Bro. The project was renamed Zeek in 2018, and it remains open source, supported by a community of research and operational users.

The Logs Are the Product

LogContents
conn.logEvery connection: endpoints, ports, duration, bytes each way, and state
dns.logQueries, response codes, and answers
http.logMethod, host, URI, user agent, status, and referrer
ssl.log and x509.logServer name, cipher, validation result, and certificate details
files.logFiles observed in transit, with hashes and MIME types
smb, kerberos, ssh, dhcpProtocol-specific records that make internal activity legible
notice.logEvents the analyst’s scripts decided were worth highlighting

Each line carries a connection identifier that ties records across logs together, so an analyst can follow one session from the TLS handshake to the file that crossed it. The Community ID field links Zeek records to those produced by Suricata, which is how teams correlate alerts with context.

Why This Shape of Data Wins Investigations

Signature engines answer a question you asked in advance. Zeek answers questions you did not know you would need to ask. When an indicator arrives three weeks after the fact, the useful query is whether any host in the estate contacted that domain or presented that certificate, at any point in the retention window. Logs answer that in seconds; stored alerts cannot, because no alert was ever generated.

This was the pattern during the SolarWinds investigation at the end of 2020. Organisations with retained network records could search historical DNS and TLS data for the published indicators and establish whether they had been affected. Those relying on alerts alone had nothing to search.

Volume is manageable too. Full packet capture at any scale is expensive to store; Zeek logs are a fraction of the size and retain most of the investigative value, which is why retention periods of six to twelve months are realistic.

Scripting and Detection

Zeek includes an event-driven scripting language, and much of its behaviour is written in it. Scripts respond to events such as a new connection, a completed HTTP request, or a certificate being seen, and can maintain state across a session or across hosts. Detections expressed this way describe behaviour rather than content: a host contacting many destinations on one port within a short window, a certificate that fails validation on an internal service, a file hash seen for the first time anywhere in the estate.

The Intelligence Framework complements this by matching observed addresses, domains, hashes, and certificate fingerprints against indicator feeds in real time, writing hits to a dedicated log.

Running It

Zeek deploys on a tap or span port and is passive, so it cannot block and cannot be the cause of an outage. At scale it runs as a cluster, with worker processes handling traffic shares, a manager aggregating, and a logger writing output. JSON output is standard practice, because it loads into a SIEM without parsing rules.

Sensor placement determines what you learn. Perimeter sensors show what enters and leaves; internal sensors at segment boundaries are what reveal lateral movement, which is invisible from the edge. Prebuilt distributions such as Security Onion package Zeek alongside Suricata and a search interface, which is the usual starting point for a team building this capability.

What It Does Not Do

Zeek is not a prevention control, it does not decrypt traffic, and it generates relatively few alerts by default, so it is not a replacement for a signature engine or for endpoint detection. It is also personal data: HTTP records, DNS queries, and connection logs describe what identifiable employees did, which brings UK GDPR obligations around lawful basis, retention, and access control.

Used properly, it is the layer that makes everything else investigable. Our SOC Maturation Services cover sensor placement, log retention, and building detections on this telemetry.

FAQ – Frequently Asked Questions

Is Zeek an Intrusion Detection System?

Not in the conventional sense. It is a network security monitoring platform: its primary output is structured logs describing observed activity rather than alerts matched against signatures. It can raise notices through its scripting framework and match indicator feeds, but organisations typically run it alongside a signature engine such as Suricata rather than instead of one.

How Much Storage Do Zeek Logs Need?

Far less than full packet capture, usually a few per cent of the traffic volume depending on which protocols dominate and how much is logged. That makes six to twelve months of retention realistic for most organisations, which is the point: investigations routinely need to look back weeks, long after any packet capture buffer would have been overwritten.

Can Zeek See Inside Encrypted Traffic?

No, but it records a great deal around it. The TLS log captures the requested server name, cipher suite, certificate issuer and subject, validity dates, and whether validation succeeded. Connection records show timing and volume. Together these identify beaconing, unexpected clients, and suspicious certificates without any access to the payload.

Where Should Sensors Be Placed?

At the internet perimeter first, then at internal segment boundaries, particularly around server estates, administrative networks, and any environment holding sensitive data. Perimeter-only deployment misses lateral movement entirely, which is where most of an intrusion happens. Place sensors where the traffic you need to investigate actually passes, not simply where a tap was easy to install.

Related Terms

You May Be Interested In These

What Is IAST?

Cybersecurity Terms

17 November 2026

What Is IAST?

Read More
How Does Incident Response Differ in OT?

Cybersecurity Terms

17 November 2026

How Does Incident Response Differ in OT?

Read More