Alot has been going on in my head about how I should transform my homelab from a simple physical box with just plain vSphere 5.5 on it and memory consuming applications like Microsoft Exchange and Lync to a nested ESXi environment where I can test new VMware features and technologies apposed to the Microsoft technologies I have seen for many years now.

This post will be the start of a series of posts regarding the described transformation. I will start with a high level design of my current setup and the setup I want to realize.

Because I have running workloads on my ESXi host, I have to think of a way to minimize impact and prevent downtime during the transformation. Not that I run any important applications, but it’s a good excercise to pretend there are =)

Next, there will probably be some troubleshooting involved which I will also include in this series of posts. Finally, my new setup should be running by design. Any changes to this design will be reflected in the series as well.

Now, my current setup in a nutshell:

A single ESXi host with Intel i7 processor, 64GB of RAM, 5 SATA disks with a capacity of 1TB each and finally a 100GB Intel enterprise SSD disk

(To see the full specs of my hardware, see the following post: https://snowvm.com/2013/05/14/esxi-51-u1-whitebox-adventures/ except the SSD I have bought some weeks ago)

The design above shows the simplicity of my setup. A single host, running VMs on local storage protected by RAID. I must say that running 16 VMs on this setup, despite the CPU and amount of memory has it’s drawbacks (probably due to the small amount of spindles). I wonder how this all will run when I realize the following setup, which will be my new design:

Single physical host. 4 SATA disk, each representing a datastore when a virtual ESXi VSAN host will run. Each VSAN host will have a virtual disk running on the SSD to meet VSAN requirements.

I will reconfigure the RAID controller to seperate the physical 1TB disks from eachother as VSAN will distribute data across multiple (virtual) nodes which makes the need for hardware RAID disappear. For this to happen, I need to have a place to store my VMs on. It’s about 500GB of actual VM data and my repository (single 1TB disk) used to store backups on has 800GB of free space.

This will be sufficient, but the risk of losing all data when this disk fails sounds like a bad idea (as both my VM data and VM backup data will be on the same physical disk). At least, when leaving the data sit on that disk for days. I will move my VMs to this disk at a different time so I can reconfigure my RAID controller right after and move my VMs from the repository/backup disk to minimize the risk. Expect a new post soon where I will have moved my VMs forth and back in preperation for my super-duper-blazing-speed-VSAN-lab!

Thanks for reading!