VMware Cloud Foundation (VCF) 9.0 supports two types of deployment mode for the applicable components (e.g. NSX Manager, VCF Operations, VCF Automation, etc), Simple (Single-node) or High Availability (Three-node).
For Lab/Learning Purposes or Proof-of-Concepts, the Simple VCF deployment will allow users to experience the full capabilities of VCF 9.0, while minimizing the resources required when compared to VCF 5.x where you had to have three NSX Managers as an example.
Since the GA of VCF 9.0, I know many folks have started to prepare their lab environments and I wanted to share the minimal resources that you would need to deploy the initial VCF 9.0 stack also referred to as a VCF Fleet for lab purposes:
Component | vCPU | Memory | Storage Provisioned | Storage Used |
---|---|---|---|---|
1 x vCenter Server | 4 | 21GB | 642GB | ~30GB |
1 x NSX Manager | 6 | 24GB | 300GB | ~14GB |
1 x SDDC Manager | 4 | 16GB | 931GB | ~77GB |
1 x VCF Operations | 4 | 16GB | 290GB | ~13GB |
1 x VCF Operations Fleet Manager | 4 | 12GB | 206GB | ~117GB |
1 x VCF Operations Collector | 4 | 16GB | 280GB | ~11GB |
1 x VCF Automation | 24 | 96GB | 626GB | ~72GB |
Total Required vCPU: 48
Total Required Memory: 194
Total Required Storage: 3.2TB
Total Used Storage: 330GB
One major take away from the table above is the vCPU column, notice the largest vCPU configuration is for VCF Automation (VCFA) which requires a 24 vCPU VM. To be able to deploy the complete VCF 9.0 stack including VCFA, you will need to have a physical ESXi host that is capable of provisioning 24 vCPU VM, that means you MUST have a physical host that has at least 12 Cores / 24 Threads to successfully deploy VCFA.
Note (06/23/25): If you can not meet the 24 vCPU minimum for VCFA, it has been shared by Tomas Fojta, that you can adjust the VCFA VM by manually editing the VM in the vSphere UI and adjust it down to 16 and potentially even 12, YMMV on overall performance and function beyond the initial setup. There are no workarounds or hacks for the memory requirements as these resources will be used by VCFA.
Here are two examples of consumer systems that you may come across:
- The Minisforum MS-A2 (which works fine with ESXi 9.0) comes with 16 Cores / 32 Threads and is capable of running VCFA
- The GMKtec NucBox K11 (which also works fine with ESXi 9.0) comes with 8 Cores / 16 Threads will NOT be capable of running VCFA
Note: I will have a through write-up on the MS-A2 and my own experiences with running VCF 9
For production-grade hardware that is on the Broadcom Compatibility Guide (BCG), this is typically not an issue but for those using consumer grade hardware, many of the previous generation of hardware will typically max out at 8 core / 16 thread and will NOT meet the requirements for deploying the full VCF 9.0 solution.
This is also a good time to think about other capabilities that you would like to explore as part of VCF 9.0 including vSAN ESA, vSphere Kubernetes Service (VKS), NSX Virtual Private Cloud (VPC) and VCFA to name a few and ensure that you have modern hardware that will allow you to invest in your skils.
What's the minimum requirements for VVF deployment?
VVF is just vSphere (ESXi, vCenter) and VCF Operations
The reqs for VCF Automation are the killer. If a hack can be found for that that’d be great.
It’s definitely something I’ve shared w/PM but I’ve certainly asked Engr and this is currently the minimal footprint
As a Broadcom TAM I need a VCF 9 platform I can use for experience and testing. I look forward to your Minisforum MS-A2 review. In the meantime is there any other home lab platforms I should also be considering for VCF 9?
The level of software inefficiency these days is absolutely insane.
In the broadcom learning portal I do not see courses on-demand for vcf 9 yet. Still the 5.2 something is there. When would it be available any ideas?
Hi Sumit, updated training classes of new major product versions always took their time. Dont know any details, but I personally would not expect anything being released before Explore in Vegas (end of August). And On Demand video trainings were always released even later than ILT classes (Instructor-led trainings).
So enjoy the summer; there is plenty of Documentation and HOLs available to get you up to speed and then you are well prepared for a training class, if desired.
I recieved a response from the learning team, it'll be ~60days after GA
i have the same question. as the available documentation is not nearly deep enough technically or am i wrong
Hello William
Ihave only one physical server can i deploy the vcf9 for a lab
Regards
Any information on how much overcommitment (vCPU:pCPU and vMEM:pMEM) is possible without running into serious performance issues?