As artificial intelligence workloads drive higher network bandwidth requirements and more complex traffic patterns, Keysight Technologies has introduced a new modular network testing platform aimed at helping organisations validate infrastructure before deployment.
The APS-ONE-400 is a 4×100 Gigabit Ethernet (100GE) platform designed for network equipment manufacturers, telecom service providers and data centre operators. The company said the system enables testing of Layer 4-7 application traffic, encrypted Transport Layer Security (TLS) traffic, Zero Trust Network Access (ZTNA) environments and large “Elephant Flow” data transfers associated with AI and large language model (LLM) workloads.
The launch comes as enterprises and infrastructure providers face increasing pressure to validate network performance against rising traffic volumes, cyber threats and the adoption of post-quantum cryptography (PQC). Traditionally, these workloads have required multiple specialised testing systems, increasing laboratory complexity and infrastructure costs.
Keysight said the APS-ONE-400 consolidates these testing capabilities into a single 1RU appliance. The platform supports up to 400 Gbps of Layer 4-7 traffic generation, 380 Gbps of hardware-accelerated TLS throughput and 95 Gbps of Elephant Flow traffic. It can also integrate with the company’s existing APS-100/400GE portfolio and supports 100GE, 25GE and 10GE network interfaces.
For larger deployments, the platform can be combined with Keysight’s APS-M8400 appliance or APS-M1010 management controller to build hyperscale test environments capable of generating up to 16 Tbps of Layer 4-7 traffic and supporting up to 20 billion concurrent connections, according to the company.
“The exponential growth of AI and machine learning workloads is placing increasing demands on data centre, service provider and enterprise networks,” said Ram Periakaruppan, Vice President and General Manager, Network Test and Security Solutions, Keysight Technologies. He said the new platform is intended to emulate AI-related traffic patterns, including large data transfers and PQC-encrypted traffic, while reducing laboratory infrastructure requirements.