Image: blackhillsinfosec.com · rights & removal
Threat Hunting On The Network: An Active Search for Risks on the Wire
Reporting by Black Hills Information SecurityRead the original at blackhillsinfosec.com
Executive Summary
Facts Only
* Threat hunters search packet data for malicious activity.
* Endpoint threat hunting relies more heavily on logs than packet inspection.
* Perimeter defenses rely on static signatures that are quickly outdated by evolving adversary techniques.
* Behavior analysis allows threat hunters to understand expected protocols and traffic flows based on network baselines.
* Common data sources include NetFlow/IPFIX records, HTTP/HTTPs Proxy logs, VPN appliance logs, full packet captures, and IDS/IPS logs.
* Favorite tools for aggregation include SIEMs (Wazuh or Graylog).
* Tools for deep packet analysis include Zeek, Arkime, Wireshark, and TShark.
* Indicators include unusual port usage, DNS anomalies (e.g., DGA domains), failed authentication attempts, certificate anomalies, and TLS/SSL anomalies.
* Methodologies include crown jewel analysis, hypothesis-driven hunting, kill chain mapping, assumed breach methodology, and using frameworks like MITRE ATT&CK®.
Full Take
The contrast drawn between signature-based detection and behavioral analysis illuminates a fundamental tension in security: the race between static defense mechanisms and dynamic adversary adaptation. While perimeter controls aim to block known threats using signatures, threat hunting seeks to identify novel, low-and-slow activity by observing deviations from established norms—the network baseline. This shift signals an acknowledgment that reliance on known patterns is insufficient against adaptive adversaries; understanding fundamental network behaviors provides a resilient vantage point. The emphasis on looking at internal traffic, exemplified by the examples regarding user access protocols (SSH in unusual locations) and data exfiltration patterns, moves the focus from mere compliance checking to contextualizing operational reality for internal users.
The suggested methodologies—hypothesis-driven hunting, kill chain mapping, and assuming a breach—suggest a necessary shift from reactive incident response to proactive, intelligence-led investigation. This framework forces the analyst to operate under the assumption that systems are already compromised, demanding an adversarial mindset rather than a purely defensive posture. The underlying implication is that true security insight emerges not just from the data itself, but from applying structured, adversarial thinking to connect disparate indicators into a coherent narrative of intent and progression.
What assumptions drive the focus on internal traffic versus perimeter control? Does prioritizing behavioral anomalies over known signatures inherently shift the locus of defense from the network edge to the host and internal user behavior? Furthermore, if adversaries can successfully alter their observed network behaviors to mimic legitimate processes, how resilient is the behavioral baseline itself against subtle manipulation?
What are the unseen costs borne by the system when trust in learned baselines becomes a critical component of threat detection?
How does the necessity of assuming compromise influence organizational responses regarding internal communication and access privileges?
From the original · Black Hills Information Security
This article was originally published in the InfoSec Survival Guide: Teal Book — Threat Hunting. Read it free online HERE, or grab it on the Spearphish General Store (free digital download or a $1.25 physical copy, your call). | All threat hunters are looking for evidence of nefarious activity, hopefully before an incident happens.Read the full story at blackhillsinfosec.com
Sentinel unavailable
The automated check of the source article's wording did not complete, so no result is shown.
This looks only at the wording of the original source article, not at this page's AI-written sections. A small local AI model made this estimate. It has not been checked against known human and machine texts, so treat it as provisional. It cannot show who wrote an article.
