Skip to main content

I was thinking about building a lab at home to play, test and learn certain topics. But as my free time is very limited and I didn't want to ruin our power costs, I came to the idea to run a lab in the cloud.

Advantages:

  • low entry costs: you pay what you use/need
  • flexible/scalable: resources are only a few clicks away
  • ability to create different scenarios: offsite, DR, ...
  • new technologies = more knowledge

Disadvantages:

  • long term costs: as long as you're using/reserving resources you need to pay for them

Being a VMware guy I've looked for ways to deploy a vSphere environment or at least an ESXi host, but unfortunately the costs were much to high. Either I would have to go with VMware Cloud or rent a bare-metal machine to install ESXi myself; both are very costly.

Hyper-V on the other side is easier to deploy in the cloud, without spending too much; as long as the base system support Nested Virtualization. So I've decided to give it a try and went with Microsoft Azure.

Microsoft has published a guide on how to enable Hyper-V in their Azure machines: How to enable nested virtualization in an Azure VM
The following overview shows which compute types/SKUs support nested virtualization: Azure compute unit

While computing isn't that expensive, storage can massively increase the costs of such a lab. Storage is paid per allocation, so as long as you consume the disk, you're paying for it. And the prices are also a bit high (from my point of view); for example a 128GiB "HDD Standard Managed Disk" costs about 5,88$/month. So this it the part where I only configured minimal resources and removed them if not needed.

All together the first system (Hyper-V + Veeam Server) didn't take much time to setup. I've started with a "D4s" machine with 4 vCPUs and 16GB RAM, which was sufficient for my purposes and affordable (0,24$/hour). There's nothing special in setting up the Hyper-V, as it works out of the box. Special care has to be taken when setting up the network; I somehow managed to kill the network of my server and had to rebuild it (easier then debugging :sweat_smile: ).
I have installed Veeam, deployed a small VM and was able to start my first backup in about an hour. I later added a second Hyper-V as a replication target and some more/faster disks to play with.

Conclusion

It was very cool to build this lab, especially because of the use of Nested Virtualization. I always wanted to increase my cloud skills, so this was/is the perfect opportunity. If you want to try out something new or create a certain scenario, the cloud is the perfect place.
On the other hand, if you want to run the lab 24/7 with a certain amount of storage, the cloud will cost too much.

At next I will try to decrease the storage costs or look for better alternatives. I also want to look at automation; that way I could automatically deploy the lab with Hyper-V, Veeam and VMs if needed and destroy it afterwards.

Yes, this is my experience, too. The prices per unit are minimal in most public clouds. But the units are small and the prices sum up to big amounts of money…

So I am looking for some small lab at home. I think for most test cases you don't need an ESXi oder vCenter. Perhaps a PC with a little bit more CPU, RAM and Disk than a standard maschine to run some hyper-v virtual systems  is enough. Sadly VMWare Server is not available anymore, for a lab this was ok…

 

It will be a mini lab, but for most tests it is ok… 


Try digitalocean as it's one of cheapest and reliable. 

 


The computing looks affordable at DigitalOcean but storage will be more expensive; 0,10$/GB. Although I haven't found a price list for their storage, perhaps they have different options.


Comment