Nested virtualization -- the ability to run a hypervisor on a virtual server -- has been largely limited to lab...
environments, but with big-name cloud providers beginning to use it, it's ready to spread to real-world contexts. Since VMware's release of vMotion in 2003, virtualization has been on an explosive growth trajectory; in its wake, this growth has fueled hypervisor development, network and storage virtualization, as well as a surrounding ecosystem.
Nested virtualization has languished in comparison, limited to pokes and prods in lab environments with scarce hardware and nonexistent performance requirements. Despite it seeming counterproductive, there are actually good reasons to run a VM inside of a VM and, as adoption grows, so does the recognition of its many use cases. Due to recent innovations and market development with Google Cloud, nested virtualization is now poised to step outside of the lab and spread like the early days of virtualization.
Nested virtualization starts in the cloud
Running a VM on top of a VM creates a hypervisor performance tax, but even with performance improvements from Intel and Advanced Micro Devices extensions, there hasn't been significant growth in nested virtualization adoption.
There are signs, however, that nested virtualization is starting to take hold in cloud computing. Google recently announced that it has brought nested virtualization to its Google Compute Engine VM instances.
The biggest benefit of nested virtualization inside a cloud provider is the improved portability of on-premises workloads. VMware dominates the on-premises virtualization market, but most cloud providers don't run the VMware hypervisor. This makes it difficult to move a VMware-based VM to a cloud provider. This is where nested virtualization comes into play.
By running nested virtualization inside the cloud provider infrastructure, IT organizations can more easily move a VM from their on-premises hypervisor to the nested virtualization hypervisor inside the cloud provider. There's no longer a need to convert the VM from one format to another, nor worry about compatibility issues with VM settings. Most significantly, the ability to place your disaster recovery (DR) site in the cloud is a huge benefit to organizations. The freedom from having to run largely unused servers, storage and hypervisors will lead to tangible benefits.
A nested virtualization use case
I was recently at the office of the hyperconverged infrastructure (HCI) provider Scale Computing. The majority of Scale Computing's customers are SMBs, for whom DR can be difficult and costly. To address these challenges, Scale Computing partnered with Google for a DR demo that used its HCI appliance and Google Cloud nested virtualization. In the demo, Scale Computing simulated a power failure with its HCI appliance, and the VMs came up automatically in Google Compute Engine. Needless to say, this was a compelling argument for any Scale Computing customer looking for a DR strategy.
Though nested virtualization has been around for a long time, performance limitations and technical complexities hampered its adoption in most environments. With more cloud providers, including Azure and Google Cloud, making it easier to use, nested virtualization is finally poised to move out of the home lab and into the real world.
Label hypervisor nesting levels to avoid confusion
Find out which hypervisors support nested virtualization
Explore different use cases for nested virtualization