← All stories
● Covered by 1 source · 1 reportLow impact1 neutral

Experimenting with ClusterNetworkPolicy in GKE for LLM Traffic Control

🔄 Updated 11d ago
New to BrevFeed? We gather this story from every outlet covering it into one summary — ranked by real-world impact, not just the latest headline — so you never miss what matters. What is BrevFeed? →

Key points

  • CNP provides cluster-wide network policy enforcement in GKE.
  • It uses a three-tier precedence: Admin, Network Policy, and Baseline.
  • Experiment focused on restricting LLM traffic and metadata API access.
  • GKE Dataplane V2 (eBPF) enforces multi-tier policies.

Introduction to ClusterNetworkPolicy

Standard Network Policies in Kubernetes are limited to namespace-level application. ClusterNetworkPolicy (CNP) extends this capability by allowing granular networking policies to be applied across an entire GKE cluster. This enables scalable security and governance policies for cluster administrators.

How CNP Functions

CNP operates with a three-tier flow, where each tier has a specific precedence. The Admin tier has the highest precedence, allowing cluster administrators to enforce mandatory security rules that cannot be bypassed, such as blocking metadata access. The Network Policy tier, with middle precedence, consists of standard namespace-scoped policies managed by application teams for controlling ingress and egress of microservices. The Baseline tier has the lowest precedence, where cluster administrators establish fallback defaults, like a default-deny posture across namespaces.

Experiment Design Pattern

An experiment using CNP aimed to restrict traffic to an LLM namespace, permit only authorized clients, and prevent cluster-wide access to the metadata API, except for DNS port 53. This setup utilized several services including Virtual Private Cloud (VPC) for private networking, TPU v5e for model inference, GKE with Dataplane V2 (eBPF) for policy enforcement, and Gemma 4 as a sensitive AI workload. CNP and standard NetworkPolicies were used for policy enforcement, with Cloud Logging capturing packet verdicts and audit trails.

Setup Overview

The environment setup involved building a custom private network infrastructure and deploying TPU-enabled clusters. This began with establishing a custom VPC with a dedicated subnet to host the necessary resources.

✨ This summary was generated by AI from the outlets' reporting listed below. It is not independently verified and may contain errors — check the original sources. How BrevFeed works →

The daily brief

One email each morning: the day's tech stories, clustered across outlets and summarized. No account needed.

One email a day. Unsubscribe in one click, any time.

Today's brief

Spend a few minutes, get the whole day. Every topic's top stories in one hands-free rundown — listen, watch, or read the transcript.

~4 min · 3 stories · Nov 21

▶ Play today's brief Listen on Spotify

New every morning, and the back catalogue is archived by date.

Reporting from

A Google Developer Relations Engineer experimented with ClusterNetworkPolicy (CNP) in Google Kubernetes Engine (GKE) to control traffic to an LLM and the cluster. CNP allows for granular, cluster-wide network policies, offering a three-tier precedence flow for security and governance. This functionality enables administrators to enforce mandatory rules and establish baseline defaults across all namespaces.