# 网络 · 日跟踪 · 2026-10-08 > 日期:2026-10-08(Asia/Shanghai,周四)。专题正文:`intel/topics/network.md`。原料:`/workspace/intel/archive/2026-10-08/`(窗口 10-07 早报后→10-08 早);对照 2026-10-07 日跟踪。抓取:Saved 276 / skipped 27 / errors 44;blocked_hosts:community.hpe.com、docs.redhat.com、e.huawei.com、support.huawei.com、www.reddit.com、www.sangfor.com(**blocksandfiles.com / www.nextplatform.com 本日恢复**;theregister 仅首页标题壳)。全库新增 URL 26 个。 ## 摘要 - **本日无公开战略级网络新稿。** - 轻(赞助内容)· Cisco「Scale-Across」:电力约束迫使单个训练作业跨多数据中心,需深缓冲路由、相干光、Silicon One、硬件 MACsec/IPsec——**NP 赞助访谈,无产品规格**。 - 旁证 · NVIDIA Sentry on BlueField-4(Agent 网络/数据访问在 DPU 带外执行,NP 09-30 稿首次入库);Dell PowerScale NFS over mTLS(2026-11)。 - 续档:VCF 9.1 SmartNIC IPsec 卸载;DGX Station 2×400G CX-8;RH Telenor LBO UPF。 ## 原文摘录 ### NextPlatform · When One Datacenter Is No Longer Enough(10-06;**赞助内容**) - 归档:`industry/when-one-datacenter-is-no-longer-enough-579afa12.txt` - URL:https://www.nextplatform.com/ai/2026/10/06/sponsored-when-one-datacenter-is-no-longer-enough/5301045 > As clusters grow, power availability is becoming a hard physical constraint, forcing infrastructure teams to consider how a single training workload can operate across multiple data centres. > The interview covers hardware-accelerated MACsec and IPsec, as well as the role of programmability in helping infrastructure adapt as AI workloads continue to evolve. ### NextPlatform · Nvidia Wraps Security Layer Around Agentic AI To Stop Rogue Behavior(09-30,本日首次入库;媒体报道) - 归档:`industry/nvidia-wraps-security-layer-around-agentic-ai-to-stop-rogue-behavior-8db6b2b6.txt` - URL:https://www.nextplatform.com/ai/2026/09/30/nvidia-wraps-security-layer-around-agentic-ai-to-stop-rogue-behavior/5300098 > First is OpenShell, an open source runtime first announced in March that is used to run agents in sandboxed environments with isolation at the kernel level. > OpenShell also can be used with chips from Intel and AMD. > "Sentry provides in-silicon security enforcement, meaning that if an AI agent attempts to move outside its software boundary, Sentry quarantines and stops it in milliseconds," > Notably absent is OpenAI, which clearly needs to rein in its models.