Aryaka Networks Inc., a provider of secure access service edge networking platforms, today announced what it says is an industry-first cloud network that’s specifically designed to support generative artificial intelligence workloads.

The new Aryaka AI>Perform network integrates strong security and observability capabilities, and is delivered natively via the Aryaka Unified SASE-as-a-service. It provides customers with a higher level of performance to support generative AI applications and services globally and at scale, the company said.

Available to early-access customers now, ahead of general availability in the third quarter, Aryaka AI>Perform is designed to overcome what the company says are “significant networking and security challenges” around generative AI. For instance, the large language models that power those workloads require enormous amounts of data to be fed into them, putting a big strain on traditional networks. The problem, according to Aryaka, is that existing networks only have a limited capacity, which makes large-scale data movement both challenging and risky.

Aryaka explained that as generative AI initiatives evolve, most companies are looking at implementing what’s known as retrieval-augmented generation, which is a technique that involves connecting LLMs to proprietary data to enhance their knowledge.

RAG can make generative AI models much more powerful, but it also puts existing networks under enormous strain, according to Aryaka Chief Product Officer Renuka Nadkarni. He said such implementations will raise the bar in terms of the converged networking and security requirements demanded by LLMs.

An AI-native network

Aryaka AI>Perform is meant to meet these higher requirements. It’s built atop of Aryaka’s recently launched Unified SASE-as-a-Service platform, which is a new, more secure networking solution focused on performance, agility, simplicity and security. The Unified SASE-as-a-Service offering bundles Arkaya’s single-pass architecture, global private network backbone, plus security, observability and application performance monitoring expertise, delivering a network service that can scale to handle the most demanding application workloads.

Nadkarni said the company’s Unified SASE-as-a-Service product is ideal for accelerating AI initiatives. “By leveraging Aryaka’s single platform with Aryaka AI>Perform, organizations can unlock the full potential of AI workloads and applications without compromising on performance and reliability,” he said.

Designed to support high-performance AI workloads globally, AI>Perform leverages the company’s core global networking infrastructure, Aryaka Zero Trust WAN, as a private, low-latency backbone. Through this, it can offer more predictable performance for generative AI workloads, ensuring data can be fed into LLMs rapidly, no matter where the customer is located.

It boasts a global reach, the company adds, leveraging its global points of presence and optimized network routes to ensure AI can be accessed from anywhere. Meanwhile, the OnePASS single-pass architecture helps to ensure the network can scale as required, without customers needing to worry about adding more infrastructure.

Perhaps the biggest advantage is the simplified management offered by AI>Perform. Users can manage everything via a centralized portal that provides full visibility and control over the network and its security, with tools available for performance monitoring and troubleshooting.

Looking forward, Aryaka said it won’t stand still once its dedicated generative AI network launches. It has plans to boost its performance with additional capabilities to be added soon.

Image: SiliconANGLE/Microsoft Designer