Future-proof Kubernetes connectivity for AI
NGINX Gateway Fabric, part of F5 NGINX One, delivers app, API, and model-aware routing with built-in governance and security.
Built for AI-ready Kubernetes connectivity
Model-aware routing for AI workloads
AI inference workloads are unpredictable—prompts vary in size and compute intensity, making traditional routing inefficient. Static load balancing can overload backends or underutilize resources. NGINX Gateway Fabric enables model-aware routing, directing traffic based on model type, cost/performance profiles, and runtime signals for efficient, scalable AI delivery.
- Route by model and workload type: Direct requests based on model type, version, or other parameters, ensuring optimal backend selection and efficient utilization of resources.
- Handle variable AI workloads: Adapt dynamically to fluctuating prompt sizes and compute demands, avoiding overload while maintaining consistent performance.
- Improve performance and efficiency: Optimize latency and throughput by routing requests to the most suitable backends based on real-time metrics such as cache and queue depth.
NGF Integration with F5 AI Guardrails and F5 WAF
- Advanced AI Threat Mitigation: Inspects and filters incoming traffic to block prompt injection and jailbreak attacks directly at the cluster edge before they reach your large language models.
- Sensitive Data Protection: Enforces strict outbound policies to prevent the exfiltration of Personally Identifiable Information (PII) and intellectual property in model responses.
- Integrated WAF Protection: Applies enterprise-grade Web Application Firewall (WAF) protection using developer-friendly GitOps workflows for native threat prevention.
- Declarative Compliance: Enables security teams to author global compliance and protection policies while application teams safely deploy them via Kubernetes-native custom resources.
Expressive traffic control
Traditional ingress relies heavily on annotations, leading to fragmented and hard-to-manage configurations. NGINX Gateway Fabric introduces expressive, Kubernetes-native CRDs that enable consistent, reusable, and declarative traffic policies across teams and environments.
- CRDs over annotations: Replace scattered annotations with structured, Kubernetes-native APIs that improve readability, consistency, and long-term maintainability.
- Reusable traffic policies: Define routing and traffic policies once and reuse them across environments, reducing duplication and configuration drift.
- Declarative configuration model: Manage traffic using version-controlled, declarative resources that align with Kubernetes best practices.
Role-based control for platform teams
As teams scale, managing shared infrastructure becomes complex and error prone. NGINX Gateway Fabric introduces role-oriented resources that separate platform and application concerns, enabling safe delegation while maintaining centralized governance and control.
- Clear separation of roles: Allow platform teams to define infrastructure and policies while developers control application-specific routing safely.
- Safe self-service enablement: Enable teams to deploy and manage services independently within guardrails, reducing bottlenecks and risk.
- Consistent governance at scale: Apply centralized policies across clusters and teams while maintaining flexibility for individual applications.
Evolve beyond ingress without disruption
NGINX Gateway Fabric provides a forward-looking model for traffic management, enabling teams to modernize Kubernetes connectivity while building on existing ingress patterns and operational familiarity.
- Next-generation ingress model: Adopt a more flexible and extensible approach to traffic management designed for modern applications and evolving workloads.
- Built for modern workloads: Support apps, APIs, and AI use cases with advanced routing, policy control, and scalability.
- Flexible adoption: Run Ingress and Gateway API side by side—use Gateway API for new apps while maintaining existing Ingress workflows without disruption.
GatewayLink - Extending BIG-IP to Kubernetes
- Eliminates Configuration Churn: Shields external BIG-IP hardware from the constant IP changes of dynamic Kubernetes pods by providing a single, highly stable NGINX endpoint.
- Bridges NetOps and DevOps: NetOps maintains secure control at the BIG-IP edge, while DevOps retains full agility for internal application routing.
- Optimized Traffic Flow: Streamlines north-south traffic from the network edge deep into the Kubernetes cluster, reducing latency and simplifying multi-cluster routing architectures.
Product overview
Efficient Kubernetes governance
NGINX Gateway Fabric delivers applications in Kubernetes using the Gateway API to manage traffic at scale. It enables platform teams to control access while application teams define routing, ensuring secure and consistent delivery to in-cluster services.
NGINX Gateway Fabric product documentationCore Capabilities
NGINX Gateway Fabric delivers Gateway API with advanced routing, policy control, and model-aware traffic management, enhanced by NGINX Plus features for observability, security, and reliability.
Advanced L7 routing
Route traffic by host, path, headers, and methods.
Traffic splitting
Enable canary releases with weighted routing.
Model-aware routing
Route AI traffic by model type, version, or cost/performance profile.
Policy-based governance
Apply reusable traffic and security policies.
Role-based access control
Separate platform and app-level responsibilities.
Real-time observability
Monitor traffic with metrics and live insights.
Dynamic reconfiguration
Update routing instantly without reloads.
Built-in security controls
Enforce TLS, mTLS, and authentication policies.
Resources
FEATURED
NGINX Gateway Fabric Supports the Gateway API Inference Extension
NGINX Gateway Fabric supports Inference extension for Gateway API to enable smart, inference-aware routing for Kubernetes
Read the blog