# VMware / Broadcom 日跟踪 · 2026-10-06
> 日期：2026-10-06（Asia/Shanghai，周二）。原料：archive/2026-10-06/vmware/。窗口：10-05（周一）早报之后 → 10-06（周二）早。上一日成功日跟踪 2026-10-05。抓取全库：主跑 Saved 267 / skipped 28 / errors 40；Scale Computing 源修复后定向补抓 Saved 14 / skipped 6 / errors 0；blocked：blocksandfiles.com / docs.redhat.com / e.huawei.com / support.huawei.com / www.nextplatform.com / www.reddit.com / www.theregister.com。

## 摘要
- **新稿 · Optimizing AI Deployments with VCF（Part 3：Maximizing CPU and Memory Utilization for Best TCO）**（VCF 博客 10-05；Broadcom Performance 团队四篇系列第 3 篇）：VCF 9.1 + MLPerf Inference 5.1 Llama3.1 8B（Dell XE7745、2×EPYC 9555、4×RTX 6000 Blackwell 96GB）测试——vCPU 16→128、内存 32→512 GB 对 GPU 推理吞吐/时延"little or no impact"；Llama3 70B（TP=4，4×H200）与 HammerDB TPC-C（1.8M TPM、DB VM 12 vCPU 75%）同机"no measurable impact"；主张把 GPU 推理 VM 多余 CPU/内存让给 Agentic AI / 数据库——**厂商性能团队自测，非第三方基准；与 09-28 Part 2（GPU 利用率；本地 09-29 首见）同系列续写**。
- **新稿 · Forrester Spotlight Finds That VCF Streamlines Private Cloud Operations**（VCF 博客 10-05）：Forrester Consulting **受 Broadcom 委托**的 2026-10 Spotlight（基于 2026-08 Projected TEI of VCF 9）——**90%** 消除第三方工具成本、**70%** 生命周期操作更快、**33%** 基础设施运维工作量下降、**50%** 工具蔓延减少——**委托研究 + 受访客户口径（Projected TEI），非独立调研；数字只可作厂商主张引用**。
- **续档 · Cybersecurity Awareness Month**：排期中 10-05「VCF Security Hardening Guidance」本次抓取**未见归档**（RSS 尚未出现或滞后）——**不编造**；10-08 首场 webinar 仍为排期。
- **轻 · Managing VKS Clusters at Scale（09-30）** 页面措辞微调（补全 VCF/VKS 全称），**无实质增量**。
- **续档**：VCF 9.1 Express Patching、SmartNIC IPsec 卸载（9.1）、memtier、HCX 9.1 Part 3、双 MQ Leaders PR；VR 09-23 MQ 媒体转述仍有效、**本地无 Gartner PDF**。

## 原文摘录

### [Optimizing AI Deployments with VMware Cloud Foundation (Part 3)](https://blogs.vmware.com/cloud-foundation/2026/10/05/optimizing-ai-deployments-with-vmware-cloud-foundation-part-3/) · 新稿（厂商性能博客）
- 归档：`vmware/optimizing-ai-deployments-with-vmware-cloud-foundation-part-3-f4be665b.txt`
> Even when we allocated only 25%-50% of available CPU cores to the VM, the inference performance is on par with bare metal.
> We ran Llama3.1 8B of MLPerf Inference 5.1 (Interactive mode) in VCF 9.1 on the system with a Dell PowerEdge XE7745 with two AMD EPYC 9555 64-core processors, 256 total logical cores, 768 GB memory, and four NVIDIA RTX 6000 Blackwell 96 GB GPUs.
> … changing the number of vCPUs while fixing the memory size, and changing the memory size while fixing the number of vCPUs, had little or no impact on the performance of AI workloads that use GPUs for acceleration.
> The HammerDB workload runs at 1.8 million TPM, which represents a heavy enterprise-level load. The database server VM runs at 75% utilization on 12 vCPUs. The LLM-inference uses 1 vCPU. The results show that neither workload experiences a measurable performance impact.
> In the future, we will demonstrate with agentic workloads.（**厂商自测；Part 4 time-slicing 待发**）

### [Forrester Spotlight Finds That VMware Cloud Foundation Streamlines Private Cloud Operations](https://blogs.vmware.com/cloud-foundation/2026/10/05/forrester-spotlight-finds-that-vmware-cloud-foundation-streamlines-private-cloud-operations/) · 新稿（委托研究推广）
- 归档：`vmware/forrester-spotlight-finds-that-vmware-cloud-foundation-streamlines-pri-4533d242.txt`
> - 90%: Eliminated third-party tool costs following consolidation onto VCF 9.
> - 70%: Faster lifecycle operations to reduce maintenance windows.
> - 33%: Reduction in infrastructure operations effort.
> - 50%: Reduction in tool sprawl yielding savings through software consolidation and lower vendor management costs.
> Source: Results are based on interviewed customers. Forrester: VMware Cloud Foundation 9 Spotlight: Streamlining Private Cloud Operations, a commissioned study conducted by Forrester Consulting on behalf of Broadcom, October 2026.（**委托研究**）

## 对基础档案的影响
- **最近动态追加**：AI 系列 Part 3（GPU 推理 VM 与 CPU 型负载/数据库同机共置，厂商自测）+ Forrester 委托 Spotlight（90%/70%/33%/50%，受访客户口径）——**用于识别 VMware「VCF=AI 私有云 + 运维整合 TCO」话术，不作独立证据**。