OpenStack Part 2: Virtual machines

OpenStack is a Cloud Framework, it doesn’t define the system, it is just a collection of cloud tools that integrate together into whatever cloud system you need. As the OpenStack Architecture Design Guide suggests, there are a multitude of possibilities.

My idea is to go through that list one by one, unlocking more services and growing my little cloud as I go further. First up is a general purpose cloud. This includes the most basic OpenStack components, and should allow me to launch virtual machines, configure some basic networking, file and object storage, etc. This is basically an IaaS (Infrastructure as a Service) model.

At this point my idea is to run OpenStack in different VMs on my physical machine. This way, if something goes wrong and I mess up the installation too much, I can just restore a VM and start over. Additionally I can also create compute nodes at will (as far as the HW will allow it).
This means I’m going to need nested KVMs.
This short guide told me how to enable it. More information can be found here.

This is what you need to end up with:
$ cat /sys/module/kvm_intel/parameters/nested
Y

Creating virtual machines: https://wiki.debian.org/KVM
$ sudo apt-get install qemu-kvm libvirt-bin
$ sudo adduser panic1 kvm
$ sudo adduser panic1 libvirtd
$ echo "LIBVIRT_DEFAULT_URI=\"qemu:///system\"" >> .profile
$ sudo vim /etc/default/libvirt-guests

From this point forward I used the virt-manager application to create a VM (make sure you select to copy the host cpu information), and install a minimal debian or ubuntu on it. Repeat the same steps needed to enable nested VMs.

The same can ofcourse be achieved just the same with VMs in VirtualBox, running on a Windows machine.

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.