Skip to content

How to Clean Up Overly Permissive Firewall Rules

Overly permissive rules are safest to clean up based on actual traffic. NxPortrait identifies unused and non-compliant rules. For overly permissive and unattributed rules, it receives traffic logs in a distributed manner, builds optimization proposals with machine learning, and then automatically places optimized rules above the overly permissive rules. There is no limit on the profiling period.

Analysis alone does not clean anything up

Even with a report analyzing overly permissive rules, the rules stay in place if the optimization plan cannot be applied to the actual firewalls. NxPortrait does not stop at analysis: it builds the justification for each task and applies the optimized rules automatically (automatic provisioning).

A six-step cleanup process

  1. Collect policies and logs: Policies from Korean and global multi-vendor firewalls and cloud firewalls are synchronized in real time, and traffic, audit, and change logs are received in a distributed manner (200Gib+ per hour, clustering & sharding).
  2. Identify unused and vulnerable rules: Log-based unused rule analysis, policy delta comparison, compliance risk analysis, and correlation analysis between rules find the cleanup targets.
  3. Profile traffic: Rule-matched and unattributed logs, and UDP and TCP 3-way handshakes, are distinguished to understand the traffic actually in use. There is no limit on the profiling period.
  4. Generate machine learning optimization proposals: Rules are split into Pareto & long tail by hit and session distribution, and distance- and density-based unsupervised machine learning is applied. The rule optimizer optimizes the group, range, and subnet objects within each rule.
  5. Deploy and verify automatically: Optimized rules are placed above the overly permissive rules automatically, and policies are re-collected after deployment to confirm the result. The justification and audit evidence are recorded together.
  6. Report: Results are shared through on-screen analysis tools, Excel and PDF reports, and status visualization boards.

Rules worth cleaning up at the same time

  • Unused rules: Rules that are not in use are found from logs and removed.
  • Non-compliant rules: Compliance and risk are checked at each stage: request, design, and physical policy.
  • Unattributed rules: Traffic that cannot be matched to a specific rule is also profiled and included in optimization.

For technical details, see Technology > ML-Based Optimization & Automated Deployment for Overly Permissive Rules.

Contact Buy via PPS