Fortinet and Arista combine best-of-breed networking and security for AI data centers

"Fortinet and Arista Networks are joining forces to take on the architectures of Cisco and HPE. Together, they offer the Fortinet Secure AI Data Center solution, which delivers a zero-trust architecture with the scalability and validation that competitors also guarantee. However, the big difference with Cisco and HPE is the promise of more freedom of choice and, with that, a best-of-breed policy."

"In addition to the claim that Fortinet and Arista jointly deliver the required scale and reliability, Fortinet Secure AI Data Center revolves around its modular nature and performance. For example, HTTPS and TLS are offloaded to Fortinet's ASIC, with 33 times the performance compared to CPU-driven processing. Those CPUs can therefore focus on their contribution to AI workloads, such as forward passes and token sampling. This keeps network traffic calmer, faster, and less variable (jitter)."

"Although GPUs or TPUs play a crucial role in this, they can never achieve their theoretical speed if the network does not provide them with data in a timely manner. This is where Arista Networks' low latency comes to the rescue for users, in addition to load balancing for clusters in the world's largest data centers. Fortinet, in turn, provides the firewalls, threat detection, zero-trust segmentation, encrypted traffic inspections, and automated responses to AI workl"

Fortinet and Arista deliver a combined Secure AI Data Center that pairs Fortinet's security capabilities with Arista's low-latency networking to compete with Cisco and HPE. The design emphasizes modularity, performance, and best-of-breed choice across vendors. Fortinet offloads HTTPS and TLS to its ASIC with 33× CPU performance, freeing CPUs for AI tasks like forward passes and token sampling, reducing jitter. Arista supplies low-latency switching and cluster load balancing for hyperscaler-scale deployments. Fortinet contributes firewalls, threat detection, zero-trust segmentation, encrypted-traffic inspection, and automated responses, enabling scalable, validated zero-trust architecture for high tokens-per-second AI workloads.

Read at Techzine Global