> ## 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. # The management plane got smaller. Your estate did not. - URL: https://www.mohammadsiddiqui.com/the-management-plane-got-smaller-your-estate-did-not/ - Published: 2026-10-25T23:27:13.000Z - Updated: 2026-10-25T23:27:13.000Z - Description: VCF 9.1.1 right-sizes VCF Management Services and adds a Small HA option, but an upgraded estate keeps its old sizing until somebody runs KB 455842. What that does to the with-or-without-the-management-plane decision. - Author: Mohammad Siddiqui - Tags: architecture, operations The objection I hear most often about VCF 9 is not about NSX, or vSAN, or the licensing. It is about the management plane. Somebody works out what VCF Management Services costs them in hosts before a single workload runs, and the conversation changes tone. It is a fair objection. On a small estate the management footprint can be a meaningful share of the cluster, and "pay this much to run the thing that runs the thing" is a hard sell to a CFO who has already been through one price conversation this year. VCF 9.1.1 does something about it. Not as much as the objection deserves, but something. And the way it has been delivered means most estates will not get the benefit unless somebody goes and takes it. ## What actually changed VCF 9.1.1 right-sizes the Management Services components for a Day 0 deployment, which is the state most environments are in: the platform is there, but log management and real-time metrics have not been turned on. The figure that gets quoted is one fewer worker node in a Simple, non-HA deployment. [William Lam](https://williamlam.com/2026/09/vcf-9-1-1-adopting-the-reduced-vcf-management-services-vcfms-footprint-after-an-upgrade.html?ref=mohammadsiddiqui.com) puts a number on that node: 12 vCPU and 24 GB of memory. A three-worker deployment becomes two. Be careful with the percentages You will see larger reduction figures quoted around this release. I have not found a before-and-after total for vCPU, memory and storage that Broadcom states directly, so I am not going to repeat one. The node count and the node size are documented. Build your business case on those and leave the percentage out of the steering pack. There is a second change that gets less attention and is arguably the more useful of the two. Small HA is a new deployment option for VCF Management Services: high availability without having to step up to the next size profile. That matters because of how the sizing conversation used to go. You wanted resilience on the management plane, resilience meant the next profile up, and the next profile up meant a conversation about hosts. Small HA removes that step. In the Installer API it is a deployment size value of `small_ha` in the cluster spec, and you can also scale a Small non-HA deployment to Small HA afterwards, under Build, then Lifecycle, VCF Management, Components, VCF Services Runtime, Action, Scale. One caveat worth carrying into a design review: VCF Automation does not have a Small HA form factor. It deploys as a single appliance. If your availability requirement covers automation as well as operations, Small HA does not finish the job for you. ## The part that catches existing estates Here is the bit to pay attention to, because it is the difference between a benefit on a slide and a benefit in your cluster. The right-sizing is applied automatically to new VCF 9.1.1 deployments. It is not applied to deployments that were upgraded to 9.1.1\. Upgrade an existing 9.1 environment and your worker nodes stay exactly the size they were, running the exact resources they were running, while the release notes tell everybody the footprint has come down. This is the whole point of the article If you upgrade to 9.1.1 and do nothing else, you have paid the cost of the upgrade and received none of the footprint benefit. Nobody is going to tell you. The environment will simply carry on using what it used before. To close the gap, Broadcom has published [KB 455842](https://knowledge.broadcom.com/external/article/455842/reducing-footprint-of-vcf-management-ser.html?ref=mohammadsiddiqui.com), which carries a script called `rightsize-day0-workers.sh`. It brings an upgraded deployment to the same end state as a fresh 9.1.1 install. ### What the script does, and when it refuses It runs on a control-plane node, reached over SSH as the vmware-system-user account and then elevated. It reads the deployment’s own state rather than asking you to describe it, pulling the size profile, the HA setting and whether this is a fleet or instance deployment from the `vmsp-platform` package deployment, then applies the matching worker size, machine type and minimum replica count. Fleet deployments carry a higher minimum replica count than instance deployments, which is the sort of detail that would have bitten somebody doing this by hand. It also refuses to run in several cases, and the refusals are sensible: Log Management installed Script aborts with no changes. The Day 0 sizing assumes those components are absent. Real-time Metrics installed Same. If you have turned either of these on, the reduced footprint is not for you. Consumption cluster type Aborts. It is intended for the management cluster. VVF environments Out of scope entirely. This is a VCF Management Services change. The operational shape of it is a rolling replacement. New worker nodes are created with the optimised machine type and the old ones are removed, one at a time. On a larger cluster that takes a while, and the script runs synchronously and waits for the rollout to finish, so whoever starts it is sitting there until it is done. Treat it as a change, not a tidy-up Nodes being replaced underneath a running management plane is a change window conversation, not something to slip in on a Friday afternoon because the ticket said "resize". Verify afterwards with `kubectl get pd vmsp-platform -n vmsp-platform` and wait for the status to read Successful before you call it done. ## What this does to the with-or-without decision The question underneath all of this is the one I opened with. Do we take the full management plane, or do we stay on something lighter and manage the estate the way we always have. The honest answer before 9.1.1 was that the overhead was real and you had to want what it gave you. That has not reversed. It has moved, and it has moved in a direction that makes the decision less lopsided on smaller estates, which is exactly where the objection was loudest. What I would actually put in front of an architecture forum: New build, small estate Run the numbers again. Small HA plus Day 0 right-sizing is a different proposition to the one that was costed six months ago, and if the earlier answer was no on footprint grounds, the earlier answer is now out of date. Existing 9.1, Day 0 only You have money sitting on the floor. Schedule KB 455842 into a change window and take it. Existing 9.1, Day N components on The script will refuse, and it should. The right conversation is whether you are actually using log management and real-time metrics, or whether they were switched on during a proof of concept and never switched off. VVF today, considering VCF None of this applies to your current estate, and the right-sizing is not a reason to move on its own. It removes a reason not to, which is a different thing. There is one more number worth having in your head before you design any of this, and it comes from community rather than from Broadcom, so check it against your own build: [VCF 9.1 Management Services is described as needing 12 to 30 IP addresses](https://digitalthoughtdisruption.com/2026/07/31/vcf-operations-fleet-management/?ref=mohammadsiddiqui.com) on the management network, with an internal service runtime range of 198.18.0.0/15 that can collide with customer addressing. I have seen a deployment held up for a fortnight over less. Get the network team the address requirement early, and ask the question about 198.18 before somebody discovers it the hard way. ## What I would check this week Four things, none of which take long. Find out whether your management services deployment is Day 0 or whether somebody enabled log management during the build. That single answer determines whether any of this is available to you. If you are on 9.1 and planning the move to 9.1.1, put the right-sizing script in the same change record as the upgrade rather than raising it separately in three months, when the appetite for touching the management plane again will be lower. If you have an availability requirement on the management plane and you sized up to meet it, look at Small HA and work out whether you can come back down a profile. And if you have a VCF business case that was declined on footprint grounds, the footprint it was declined on is not the current one. ## Sources - [Customer Favorite Capabilities: Accelerating Your Path to VCF 9.1.1, Broadcom, 13 October 2026](https://blogs.vmware.com/cloud-foundation/2026/10/13/customer-favorite-capabilities-accelerating-your-path-to-vcf-9-1-1/?ref=mohammadsiddiqui.com) - [Broadcom KB 455842: Reducing footprint of VCF management services cluster that is only hosting Day 0 component](https://knowledge.broadcom.com/external/article/455842/reducing-footprint-of-vcf-management-ser.html?ref=mohammadsiddiqui.com) - [William Lam: VCF 9.1.1, adopting the reduced VCF Management Services footprint after an upgrade](https://williamlam.com/2026/09/vcf-9-1-1-adopting-the-reduced-vcf-management-services-vcfms-footprint-after-an-upgrade.html?ref=mohammadsiddiqui.com) - [William Lam: 10 exciting enhancements in VMware Cloud Foundation 9.1.1, including the Small HA option](https://williamlam.com/2026/09/10-exciting-enhancements-in-vmware-cloud-foundation-9-1-1.html?ref=mohammadsiddiqui.com) - [Broadcom TechDocs: VCF Management Services models](https://techdocs.broadcom.com/us/en/vmware-cis/vcf/vcf-9-0-and-later/9-1/design/vmware-cloud-foundation-concepts/vcf-management-services-models.html?ref=mohammadsiddiqui.com) - [Digital Thought Disruption: VCF Operations fleet management, for the IP and addressing detail](https://digitalthoughtdisruption.com/2026/07/31/vcf-operations-fleet-management/?ref=mohammadsiddiqui.com) Related: [The staged path from vSphere 8 to VCF 9.1](https://www.mohammadsiddiqui.com/the-staged-path-from-vsphere-8-to-vcf-9-1/), which covers the upgrade sequencing this sits inside, and was updated this month with the confirmed lifecycle dates.