Juniper Networks launches a new solution for service providers and AI cloud providers, aimed at accelerating the deployment of AI services and simplifying network management. The solution combines network infrastructure management with automation and security for GPUaaS and AIaaS environments.
Juniper introduces the AI-Native Networking Platform. The network specialist has designed this solution tailored for companies offering AI services via the cloud. The platform helps businesses that provide AI or GPU-as-a-Service to customers by assisting in optimizing high-performance multi-tenant environments. It includes a combination of hardware and associated software tools.
Switches and Firewalls
The AI-Native Networking Platform utilizes switches from the QFX series, routers from the PTX series, and firewalls from the SRX series. These are managed through Juniper Apstra and Mist AI. Thanks to intent-based networking and automation, Juniper claims implementations can be faster and operational costs reduced. Additionally, integration with Red Hat OpenShift provides support for Kubernetes environments.
For a secure multi-tenant AI cloud environment, Juniper integrates zero trust security and EVPN VXLAN functionality. This protects AI models and data against threats. The next-generation firewall from the SRX 4700 series provides security and supports 400 Gbps interfaces with MACsec encryption.
Blueprints
Furthermore, Juniper offers Juniper Validated Designs (JVDs) as blueprints for hybrid environments with technologies from vendors such as Nvidia, AMD, Weka, and VAST Data. This is relevant for AI cloud providers seeking assistance in faster implementation of AI solutions.
The AI-Native Networking Platform offers managed service providers network automation and AI-driven management tools. Juniper aims to provide customers with the tools relevant for setting up scalable and secure networks that support AI workloads. Juniper refers, among other things, to model training and inference based on Retrieval Augmented Generation (RAG).