A naukri.com initiative
Siliconangle
2w
117
Image Credit: Siliconangle
Arista Networks expands AI performance with new capabilities
- Arista Networks Inc. has introduced new capabilities for its EOS Smart AI Suite to enhance AI cluster performance and efficiency.
- The company's Arista Cluster Load Balancing (CLB) optimizes AI workload performance with consistent, low-latency network flows.
- Arista CloudVision Universal Network Observability (CV UNO) now offers AI observability for improved troubleshooting and issue inferencing.
- Cluster Load Balancing is based on RDMA queue pairs to improve bandwidth utilization between spines and leaves in AI clusters.
- CLB resolves uneven traffic distribution and high tail latency issues by optimizing bidirectional traffic flow.
- Arista's CLB and CV UNO allow the network to impact AI performance at the application level, providing insights for network engineers to address performance issues.
- CLB delivers significant performance improvements, increasing throughput and reducing job completion times for organizations utilizing AI clusters.
- Arista's CV UNO offers end-to-end AI job visibility, unifying network data within the Arista Network Data Lake (NetDL) for real-time insights.
- EOS NetDL Streamer provides AI job monitoring, deep-dive analysis, flow visualization, and proactive issue resolution to ensure efficient AI workload execution.
- Arista's CLB is available on select platforms now, with plans for wider support by 2025, while CV UNO is currently available with AI observability enhancements expected later in 2025.
Read Full Article
7 Likes
For uninterrupted reading, download the app