We’ve put together a short screencast to show how our cloud responds to a complete hypervisor failure. If you’re not familiar with cloud architecture/what a hypervisor is, here’s quick overview: Cloud platforms are generally clusters of computers, running base operating systems and a hypervisor software layer. The hypervisor layer provides machine virtualisation, allowing the physical machine to run many independent guest operating systems which are unable to interfere with each other. Each guest operating system is given a pool of resources from the physical machine, and essentially runs unawares that it is being virtualised.
In our cloud, we have a control server which provides management, and deploys machine images onto physical hypervisor servers. These hypervisor servers then host and power its assigned virtual machines. In a traditional dedicated environment, a complete server failure could be catastrophic. However, in a cloud architected cluster, if one of the hypervisor servers fails in any way, the virtual machines hosted on that physical machine will have very limited downtime (usually just minutes), before they are automatically deployed onto a new server within the cluster.
Here is a short demo displaying these characteristics within the Dynomesh cloud: