I just bought two GMKtec NucBox G3 N100 mini PCs. I ordered one, barebones, directly from the manufacturer, and picked up the other on Amazon. The barebones machine was $99 with coupon. The Amazon purchase included 8GB RAM and a 256GB SSD, and came with Windows 11 pre-installed. The machine has a single DIMM slot and will take up to 32GB. It has an Intel-based I225-V 2.5Gb NIC, though I seem to be able to get ESXi 8.0U2 to connect at only 1Gb. Transceiver? MikroTik switch? Not sure. With the machine added to vCenter, I can migrate powered-off VMs to and from. It's not part of an EVC cluster, so vMotion of a running VM is not possible in both directions. It's a cute little box that cost about the same as two meals, and should last a lot longer.
Just today I received mini PC with N100. I added 32GB DDR5, NVMe 2TB and SATA SSD 2TB. Looks OK right? I have now this small machine in the cluster with my main Proxmox machine. And seriously: this is bloody slow machine! Only 4 CPU cores is a killer (comparing to my 16 cores AMD). No, not happy with the performance. Yes, it has 4x 2.5G NICs but with such slow CPU there is nothing much I can do - just replacement of Raspberry Pi. Eh...
@zyghom thank you for the comment! Definitely not surprised the N100 can't keep up with your AMD. These are good little CPUs but are not the most powerful. I definitely think you hit the anil on the head that these are more in the realm of Raspberry Pi type processors that can handle just a few workloads.
@@VirtualizationHowto I am still testing - as I said I have it only for few hours. No, it is not 16 core Ryzen 7, but yes, max power consumption is 3x less (or even more, I see around 25W with 2x SSD inside). Will see more in few days.
thanks a lot , I think this device proper for the remote Place such as our Home lab or workplace need to access from outside . I Don't Know how much Pressure can be tolerate and Not to fail with this mini Pc . It's Best to be reach access Place . the Great Point is you can run Esxi 8.0 in the scenario that's the Big Point.
My understanding of that message about not being able to establish a connection with the onboard TPM is a driver issue. I have that with my GMKtec as well. If you ever find a VIB to support the TPM under ESXi 8, please let us know.
The old hypervisor side of things is an absolute waste of overhead and space - it needs to be containerization a simple basic linux and then put docker and portainer on it and build what you want or with cloud now days just use a cloud instance... at home now days i just use dell optiplex micro desktops i7 with 32gb ddr4 sodimm and nvme they make perfect docker swarm container clusters and run what ever you want... especially now when you can have things as vscode and everything else you want to use in a container as well why have all the extra junk and get the best performance out of tiny rigs... if someone is really new to containers and still wants to feel like that old hypervisor layer just throw on proxmox and then you can still do containers at that level and more general vm's as well
@engineered1337 thank you for the comment and sharing your setup! I still favor running a hypervisor on top of bare metal as this abstracts the hardware from the platform, allowing you to easily do things like backup and high-availability of a container host VM. However, for many home labs, bare metal will also work well for sure. It depends on the use case and how critical the data is.
@@VirtualizationHowto As i mentioned above with docker swarm that is a HA cluster format and scalable automatically, as for backups its easier to just use a NFS share to a netapp or to a local NAS and then do you backups from there and multiple ways to do your backups, with hosting your own container registry on the NFS share. The way your looking at it is very rudermentary its still great but being a Senior cloud engineer and architect now days i see much better ways... i would suggest go have a look at the cloud native landscape it will give you alot of information and what sort of things you can now do with a more agile solution setups that you can run in your own setups.
@@VirtualizationHowto The real goal people need to be obtaining in a home setup is infrastructure as code with terraform utilising a VCS (version control system) be it gitlab, github, jenkins ect to be able to have CI/CD pipelines and autoamtion sets to just go ahead and rebuild a whole envrioment in minutes and be able to have your branch management system for pre-prod (UAT) and dev that once tested and spun up and destroyed with the infrastructure as code and proven to work can be pushed to the main branch and be your active system sets, you should be able to run any code nginx, python ect ect from the pipelines and be able to have a agile system enviroment
Awesome video Brandon !! These things are so nice ! One day ill get one and try some cool things on it !
Thanks Jason!
I just bought two GMKtec NucBox G3 N100 mini PCs. I ordered one, barebones, directly from the manufacturer, and picked up the other on Amazon. The barebones machine was $99 with coupon. The Amazon purchase included 8GB RAM and a 256GB SSD, and came with Windows 11 pre-installed. The machine has a single DIMM slot and will take up to 32GB. It has an Intel-based I225-V 2.5Gb NIC, though I seem to be able to get ESXi 8.0U2 to connect at only 1Gb. Transceiver? MikroTik switch? Not sure.
With the machine added to vCenter, I can migrate powered-off VMs to and from. It's not part of an EVC cluster, so vMotion of a running VM is not possible in both directions.
It's a cute little box that cost about the same as two meals, and should last a lot longer.
Thank you for the comment! Sign up on the forums and let's discuss: www.virtualizationhowto.com/community
Just today I received mini PC with N100. I added 32GB DDR5, NVMe 2TB and SATA SSD 2TB. Looks OK right? I have now this small machine in the cluster with my main Proxmox machine. And seriously: this is bloody slow machine! Only 4 CPU cores is a killer (comparing to my 16 cores AMD). No, not happy with the performance. Yes, it has 4x 2.5G NICs but with such slow CPU there is nothing much I can do - just replacement of Raspberry Pi. Eh...
@zyghom thank you for the comment! Definitely not surprised the N100 can't keep up with your AMD. These are good little CPUs but are not the most powerful. I definitely think you hit the anil on the head that these are more in the realm of Raspberry Pi type processors that can handle just a few workloads.
A N100 mini with 4x 2.5g nics? that's cool
@@SB-qm5wg yes, Topton N100 4x i226-V DDR5
@@VirtualizationHowto I am still testing - as I said I have it only for few hours. No, it is not 16 core Ryzen 7, but yes, max power consumption is 3x less (or even more, I see around 25W with 2x SSD inside). Will see more in few days.
I use these n95- n100 mini pcs for docker containers, they work wonderful for them
thanks a lot , I think this device proper for the remote Place such as our Home lab or workplace need to access from outside . I Don't Know how much Pressure can be tolerate and Not to fail with this mini Pc . It's Best to be reach access Place . the Great Point is you can run Esxi 8.0 in the scenario that's the Big Point.
@sohrabtehrani3466 Thank you for the comment! Sign up on the forums and share more of your thoughts here: www.virtualizationhowto.com/community
My understanding of that message about not being able to establish a connection with the onboard TPM is a driver issue. I have that with my GMKtec as well. If you ever find a VIB to support the TPM under ESXi 8, please let us know.
Thank you for the comment! Sign up on the forums and let's discuss: www.virtualizationhowto.com/community
Thanks!
The old hypervisor side of things is an absolute waste of overhead and space - it needs to be containerization a simple basic linux and then put docker and portainer on it and build what you want or with cloud now days just use a cloud instance... at home now days i just use dell optiplex micro desktops i7 with 32gb ddr4 sodimm and nvme they make perfect docker swarm container clusters and run what ever you want... especially now when you can have things as vscode and everything else you want to use in a container as well why have all the extra junk and get the best performance out of tiny rigs... if someone is really new to containers and still wants to feel like that old hypervisor layer just throw on proxmox and then you can still do containers at that level and more general vm's as well
@engineered1337 thank you for the comment and sharing your setup! I still favor running a hypervisor on top of bare metal as this abstracts the hardware from the platform, allowing you to easily do things like backup and high-availability of a container host VM. However, for many home labs, bare metal will also work well for sure. It depends on the use case and how critical the data is.
@@VirtualizationHowto As i mentioned above with docker swarm that is a HA cluster format and scalable automatically, as for backups its easier to just use a NFS share to a netapp or to a local NAS and then do you backups from there and multiple ways to do your backups, with hosting your own container registry on the NFS share. The way your looking at it is very rudermentary its still great but being a Senior cloud engineer and architect now days i see much better ways... i would suggest go have a look at the cloud native landscape it will give you alot of information and what sort of things you can now do with a more agile solution setups that you can run in your own setups.
@@VirtualizationHowto The real goal people need to be obtaining in a home setup is infrastructure as code with terraform utilising a VCS (version control system) be it gitlab, github, jenkins ect to be able to have CI/CD pipelines and autoamtion sets to just go ahead and rebuild a whole envrioment in minutes and be able to have your branch management system for pre-prod (UAT) and dev that once tested and spun up and destroyed with the infrastructure as code and proven to work can be pushed to the main branch and be your active system sets, you should be able to run any code nginx, python ect ect from the pipelines and be able to have a agile system enviroment