BIG-IP Next for Kubernetes deployed on NVIDIA BlueField DPUs

Delivering multi-tenancy, acceleration, and observability

BIG-IP Next for Kubernetes deployed on NVIDIA BlueField DPUs provides enterprises and service providers with a single control point to maximize AI Cloud infrastructure usage and accelerate AI traffic for data ingestion, model training, inference, RAG, and agentic AI.

Maximize Efficiency and Lower Costs

Maximize AI infrastructure investment and achieve lower TCO by providing high-performance traffic management and load balancing for cloud-scale AI infrastructure.

Multi-Tenancy Support for AI Cloud Providers

Enable secure Kubernetes-based multi-tenancy and network isolation for AI applications, allowing multiple tenants and workloads to efficiently share a single AI infrastructure—even down to the server level.

DPU-Driven Zero Trust Security

Integrate critical security features and zero trust architecture, including edge firewall, DDoS mitigation, API protection, intrusion prevention, encryption, and certificate management, while offloading, accelerating, and isolating these onto the DPU.

A Single Point of Control

Improved performance

Maximize infrastructure potential AI applications demand accelerated networking capabilities. BIG-IP Next for Kubernetes optimizes traffic flows to AI clusters, resulting in more efficient use of GPU resources by interfacing directly with front-end networks. For multi-billion-parameter AI models, BIG-IP Next for Kubernetes reduces latency and provides high-performance load balancing for data ingest and incoming queries.

Scale GPUaaS

Multi-tenancy architecture turbocharges AI factories and cloud data centers for AI workloads Enables organizations to support more users on shared computing clusters while also scaling AI training and inference workloads. Accelerate AI model connection to data storage in disparate locations while significantly enhancing visibility into app performance, by utilizing advanced Kubernetes capabilities for AI workload automation and centralized policy controls.

Protect Data, Models, and Apps

Secure and streamline your AI deployments The rapid growth of APIs for AI models introduces significant security challenges. BIG-IP Next for Kubernetes automates the discovery and protection of endpoints, securing AI apps against evolving threats. By offloading network security processing from CPUs to the NVIDIA BlueField DPUs, and by leveraging their zero-trust architecture, BIG-IP Next for Kubernetes delivers fine-grained protection and ensures robust data encryption. This approach not only enhances cyber defenses but also optimizes AI data management, resulting in more secure, scalable, and efficient infrastructure for service providers and enterprises.

Optimize North-South Traffic

Get an integrated view of networking, traffic management, and security The solution meets the growing demands of AI workloads and is purpose-built for Kubernetes environments. It enhances the efficiency of north-south traffic flows and gives organizations an integrated view of networking, traffic management, and security for AI use cases like inferencing and agentic AI.

Optimize LLM Routing and Inference

Advanced LLM routing dynamically directs tasks to the most efficient models, reducing latency, improving time-to-first-byte (TTFB), and leveraging domain-specific LLMs for higher-quality outputs. NVIDIA Dynamo integration further accelerates distributed inference using cost-efficient KV caching on CPUs, minimizing reliance on GPU memory while optimizing performance. F5 bolsters security for Model Context Protocol (MCP) deployments by serving as a reverse proxy, protecting LLMs from evolving threats and ensuring adaptability to fast-changing AI protocols for scalable and secure operations.

Deploying NVIDIA Accelerated Computing at Scale

Maximize Your Investment

Performance, efficiency, and security are central to the success of organizations deploying large-scale GPU clusters in their AI factories and cloud data centers. BIG-IP Next for Kubernetes leverages the NVIDIA BlueField-3 DPU platforms, releasing valuable CPU cycles for revenue-generating applications. BIG-IP Next for Kubernetes deployed on NVIDIA BlueField-3 DPUs (B3220 and B3240 versions) optimizes data movement and improves GPU utilization while optimizing energy consumption.

Core Capabilities

BIG-IP Next for Kubernetes delivers high-performance networking, advanced security, and simplified operations for AI factories, enabling seamless scaling, Kubernetes integration, and real-time traffic visibility to optimize AI workloads.

Maximize AI Performance

Boost data throughput and GPU utilization for AI workloads.

Secure Multi-Tenancy

Enable secure, isolated environments for multiple tenants.

Zero-Trust Security

Protect AI workloads with advanced security features.

Centralized Management

Simplify operations with a single point of control.

High-Speed Networking

Deliver ultra-fast connectivity for demanding AI tasks.

Seamless Scalability

Expand infrastructure effortlessly as AI workloads grow.

Kubernetes Integration

Integrate seamlessly with Kubernetes-native workflows.

Traffic Observability

Gain real-time insights into network traffic and performance.

Resources

FEATURED

F5 BIG-IP Next for Kubernetes delivers validated AI inference performance with NVIDIA BlueField DPUs.

Independent testing by The Tolly Group demonstrates that F5’s intelligent AI traffic management significantly improves inference efficiency, enabling higher token throughput, faster time-to-first-token, and lower latency for large-scale AI workloads running on NVIDIA-accelerated infrastructure.

Learn more