[ad_1]
Our first report highlights the new opportunities, unprecedented demand, and innovations that define the biggest transformation in Ethernet in decades.
Spirent has released its first report on the High Speed Ethernet (HSE) market and how advances in AI are impacting data centers, communications, and enterprise networking.
The impact of AI cannot be overestimated, as it will fundamentally transform data centers and interconnections beyond the impact of traditional cloud applications.
Spirent’s impact report, “The Future of Fast Ethernet in Data Center, Telecom, and Enterprise Networking,” includes insights from more than 340 HSE initiatives supported by Spirent over the past year, highlighting key drivers, market impact, and A comprehensive look at predictions about what will happen. .
“As the market focuses on the power and promise of AI, there is tremendous pressure to move faster, push the boundaries of speed, and relentlessly pursue every competitive advantage available in the market,” Spirent said. said Aniket Khosla, Vice President of Wired Product Management. .
“AI is driving an inflection point in the market, and there is a strong need to understand and stay ahead of the trends driving this.”
“As an independent vendor trusted by both service providers and network equipment manufacturers (NEMs) to validate cutting-edge Ethernet technologies and new infrastructure, Spirent has a unique perspective on the market. Our new report highlights the pace of innovation in the HSE market, the AI-driven data deluge transforming Ethernet, and the steps the industry is taking to address the resulting challenges.” said.
The most promising HSE trends identified in the report are:
HSE port shipments continue to accelerate – in 2023, suppliers shipped to more than 70 million HSE ports, with volumes expected to jump to more than 240 million ports between 2024 and 2026. Masu. Ahead of the traditional demand curve, the market is already looking to 1.6T Ethernet to pursue AI-driven opportunities next year as well. Acceleration – The impact of AI is changing data centers and the interconnection ecosystem around them, requiring networks to be redesigned to support new performance and scalability requirements. As a result, the market will continue to see a rapid migration to 400/800G and beyond. AI Fabric requires new testing approaches. AI datacenter performance testing requires test cases configured to use real servers to generate AI workloads, which is a very expensive endeavor. The result is a new, cost-effective way to stress test AI data center networking that emulates realistic xPU workload traffic.
While hyperscalers are migrating to 800G, enterprises are not waiting for future developments to advance, and carriers are abandoning traditional strategies to cater to customers in ambitious adoption cycles. Masu.
Key insights include:
The need for speed – The report notes that 800G is still gaining momentum to meet near-term needs as AI models increase in complexity and size, requiring more bandwidth and speed. We expect it to be complemented by 1.6T Ethernet soon. AI inference: Increasing capacity at the edge – Large volumes of AI traffic reside at the edge, increasing the need to upgrade the capacity of access and transport networks sooner. Early predictions are that edge locations may require additional capacity, with far-edge sites requiring speed grade upgrades of 25-50G, mid-edge sites 100-200G, and near-edge sites 400G. It has been suggested that refresh cycles could be as fast as 800G. RoCEv2 in the backend data center – Remote Direct Memory Access over Converged Ethernet (RoCEv2) is a key enabler of high-performance, low-latency networking by facilitating direct memory access between devices over Ethernet is. This report highlights the increasing adoption of RoCEv2 in backend data centers for AI interconnect fabrics.
Spirent’s impact report, “The Future of Fast Ethernet in Data Centers, Telecom, and Enterprise Networking” is available for download now.
Click below to share this article
Facebook Twitter LinkedIn Email WhatsApp
