> ## Content Index
> Fetch the complete content index at: https://www.mohammadsiddiqui.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# Two Weeks in VCF Memory Tiering’s Fine Print, the Hardware Crisis Thesis, and the Explore Session Wave
- URL: https://www.mohammadsiddiqui.com/two-weeks-in-vcf-memory-tierings-fine-print-the-hardware-crisis-thesis-and-the-explore-session-wave/
- Published: 2026-08-12T02:13:35.000Z
- Updated: 2026-08-29T04:00:03.000Z
- Author: Mohammad Siddiqui
- Tags: Newsletter, operations

*A round-up of what Broadcom published in the last two weeks that’s worth your attention: the reservation-vs-pinning distinction every memory-tiering estate needs to get right, Broadcom’s clearest public statement yet on hardware scarcity driving private cloud adoption, and the opening wave of VMware Explore 2026 session guides.*

The two weeks since the VMSA-2026-0006 patches dominated the calendar have been quieter on the release-note front and busier on the strategic and Explore-preparation front which makes sense with Las Vegas three weeks out. Three items are worth their own attention.

## 1\. Memory Tiering’s Fine Print Reservation vs. Pinning

Broadcom published a follow-up to the Memory Tiering configuration guide covering a distinction that matters the moment you run tiering in production: VM Memory Reservation and “Reserve all guest memory (All Locked)” sound similar and behave completely differently.

• VM Memory Reservation is a promise to the scheduler a guaranteed capacity floor for the VM, but the memory backing that guarantee can still participate in Memory Tiering. A reserved VM can have its reserved pages spread across Tier 0 (DRAM) and Tier 1 (NVMe); reservation guarantees quantity, not placement.

• “Reserve all guest memory (All Locked)” is a contract with the hardware, not the scheduler. Checking this box pins physical pages in DRAM at power-on, with page table mappings established immediately. Those pages cannot be moved, reclaimed, swapped, migrated, or tiered by any system component the vmx configuration sets sched.mem.pin = “TRUE”, and the hardware enforces the lock, not the scheduler.

The practical consequence: a VM with a large reservation is not automatically excluded from memory tiering, but a VM with All Locked enabled is completely excluded 100% DRAM, permanently, by design. If your tiering rollout isn’t delivering the expected consolidation gains, checking for All Locked VMs hiding in the cluster is one of the first places to look. It’s also worth an explicit design decision rather than a legacy setting inherited from pre-tiering VM templates: reservations for guaranteed capacity, All Locked reserved deliberately for the specific workloads that genuinely require hardware-enforced DRAM residency latency-critical databases, real-time processing, anything where a tiering-induced access-time variance is unacceptable.

Worth an audit line in your next memory-tiering rollout: inventory every VM with All Locked set before enabling tiering on a cluster, and confirm each one is there by design rather than by template inheritance.

## 2\. The Hardware Crisis Thesis, Stated Plainly

Broadcom’s corporate news site published a strategic piece naming the supply environment directly: a global hardware supply crunch driven by surging AI infrastructure demand, described as a “hardware super-cycle,” is producing rising costs and long lead times across enterprise IT not just for GPUs, but for the standard server, memory, and NVMe components every VCF estate depends on. The framing matches what CIOs have been reporting anecdotally all year: components once ordered on a routine cycle are now competing with hyperscaler and neocloud AI procurement at a scale that distorts availability for everyone else.

The piece’s thesis is straightforward: platforms that extract more value from existing and constrained hardware rather than assuming refresh-on-demand are the ones that keep IT organisations moving through a supply-constrained period. That’s a direct line to three things already covered in this Field Guide: NVMe memory tiering (more effective capacity from the DRAM and NVMe you can actually procure), vSAN’s ZSTD compression and global deduplication (more usable capacity from the storage you already have), and the ODM/OEM certification landscape (broader sourcing options when tier-1 lead times stretch). None of those technologies were built for a supply crunch specifically but a supply crunch is exactly the environment where their TCO case stops being optional analysis and starts being the procurement plan.

Read together with the certification carry-forward story (3,148 server models preserved from 8.x into 9.x), the strategic argument has real technical backing: the estates best positioned for a hardware-constrained 2026 are the ones that already treated their existing fleet as an asset to optimise rather than a liability to refresh.

## 3\. The Explore 2026 Session Wave

With Explore three weeks away (31 August – 3 September, The Venetian, Las Vegas), the VCF blog has shifted heavily into session-guide mode five posts in the last nine days. Rather than cover each individually, here’s the map:

• VCF Automation deep-dive positions VCF Automation as the engine behind Internal Developer Platforms, rapid application delivery, and lifecycle management, tying back to Broadcom’s recent IDC MarketScape Leader placement for private/hybrid cloud management and automation.

• New interactive workshop tracks two small-group, hands-on tracks new for this year: one on designing and architecting private cloud infrastructure, one on consumption and management across teams. Both run daily Monday–Wednesday during the event and sit alongside (not instead of) the standard 1:1 adoption workshops available through account teams. Worth booking early seats are deliberately limited for the interactive format.

• AI session track covers building and running agentic AI workflows with open-source and standards-based tooling, governing AI fleets (access scoping, observability, cost/quota controls, policy enforcement, sandboxing), and defending against AI-driven threats. The governance content pairs directly with the VKS Policy Bundle and MCP governance material already in this series.

• VCF Edge session track covers the Edge AI stack (vGPU partitioning, vLLM runtime automation, AI governance at the edge), small-footprint cluster architecture for constrained sites, and VCF Edge 9.1’s live patching and confidential computing for distributed environments relevant to retail, manufacturing, and defense estates specifically.

• Operations session guide anchored by “What’s New in VMware Cloud Foundation Operations,” covering the latest 9.1 operational capabilities for sovereign private cloud infrastructure.

If you’re planning Explore attendance for your team, this is the moment to start locking sessions the content catalogue is open now and the interactive workshop tracks in particular will fill early given the deliberately small format.

## Also Landed Already Covered Elsewhere in the Field Guide

For completeness, three technical releases from this window already have full treatment in this series and aren’t repeated here: the VKS Policy Bundle for Kubernetes admission control (Part 58), Data Services Manager 9.1 (Part 59), and the VCF 9.1 Installer/brownfield import workflow (Part 60). The vSAN hardware-refresh guidance and ODM/OEM server economics are covered in the standalone hardware economics piece.

## The Architect’s Takeaway

The through-line across this fortnight is consistent with everything else in 2026 so far: get more from what you have. Memory tiering’s reservation fine print is about not accidentally leaving capacity on the table; the hardware crisis piece is Broadcom’s clearest public argument yet for why that capacity discipline matters commercially; and the Explore session wave shows the same theme carried into governance, edge, and operations content. Audit your All Locked VMs, read the hardware crisis piece with your next refresh cycle in mind, and get your Explore sessions booked the interactive workshops won’t stay open long.

## Sources

• [Broadcom Memory Tiering and VM Memory Reservation: What You Actually Need to Know (Dave Morera, 6 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/06/memory-tiering-and-vm-memory-reservation-what-you-actually-need-to-know/?ref=mohammadsiddiqui.com)

• [Broadcom News Why the 2026 Hardware Crisis Will Accelerate Adoption of the Modern Private Cloud](https://news.broadcom.com/cloud/2026-hardware-crisis-modern-private-cloud-adoption?ref=mohammadsiddiqui.com)

• [Broadcom The Future of Cloud Infrastructure: VCF Automation at Explore 2026 (3 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/03/the-future-of-cloud-infrastructure-deep-dive-into-vmware-cloud-foundation-automation-at-explore-2026/?ref=mohammadsiddiqui.com)

• [Broadcom New Workshops at VMware Explore 2026: Design, Architect, and Consume Private Cloud (William Lam, 7 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/07/new-workshops-at-vmware-explore-2026-design-architect-and-consume-private-cloud/?ref=mohammadsiddiqui.com)

• [Broadcom VMware Explore 2026: Learn to Build, Govern, Protect and Optimize AI Workloads (Chris Wolf, 10 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/10/vmware-explore-2026-learn-to-build-govern-protect-and-optimize-ai-workloads/?ref=mohammadsiddiqui.com)

• [Broadcom VMware Explore 2026: Top VCF Edge Sessions in Las Vegas (Dhruv Tyagi, 11 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/11/vmware-explore-2026-top-vcf-edge-sessions-in-las-vegas/?ref=mohammadsiddiqui.com)

• [Broadcom Your Guide to VCF Sessions for Better Private Cloud Operations at VMware Explore 2026 (6 Aug 2026)](https://blogs.vmware.com/cloud-foundation/2026/08/06/your-guide-to-vcf-sessions-for-better-private-cloud-operations-at-vmware-explore-2026/?ref=mohammadsiddiqui.com)