The folks at Ravello Systems approached me to brief me on how they're offering "Virtualization 2.0." The exact words they used are:
With your interest in innovation the cloud, I thought you may be interested in speaking with Ravello Systems, whose founders were behind the creation of the KVM hypervisor. The movement to more complex applications and desire by enterprises to leverage the public cloud to save money has signified a fundamental shift in virtualization which we are calling the evolution to virtualization 2.0. Ravello developed of the industry’s first Cloud Application Hypervisor, and anticipated early on that it would be how virtualization needed to evolve to embrace the cloud for complex applications.
Ravello would like to discuss virtualization 2.0, specifically:
- The rise of the public cloud and complex applications and state of virtualization;
- How nested virtualization in the cloud allows for cost benefits with workload consolidation and increased performance; and
- How its binary translation technique for full virtualization is the optimal way for direct execution.
With a lead-in like that, how could I resist? So Navin Thadani, Ravello Systems' SVP of products, Alex Fishman, virtualization CTO, and Geert Jansen, director of product marketing were soon on the line with me. I spoke with a few of these folks when they were with Qumranet and developing KVM.
The folks at Ravello saw that application mobility, that is the ability to move an application from one computing environment to another, was somewhat limited by the fact that there are many different virtual machine software products in use today. VMware is offering their ESX server, Microsoft offers their Hyper-V, and the open source community is offering both Xen and KVM. Suppliers, such as Citrix and Oracle, are offering Xen-based products. Suppliers such as Red Hat, SUSE and IBM are offering KVM-based products.
Ravello saw the problem and developed technology that allowed virtual machine software products to be nested, that is a common virtual machine environment could be run inside of a number of other virtual machine software products (beginning with VMware and KVM), allowing virtual machines to be portable from one environment to another.
This, the company thought would offer a revolutionary freedom from cleverly hidden software lock-ins at the hypervisor/virtual machine software layer.
Their next step was to make it easy for a customer to encapsulate their applications, databases, and tools so that they could be easily replicated. This, Ravello hoped, would make it very easy for customers to create complex, distributed, virtualized workloads using a drag-and-drop editor.
The secret sauce was creating a common virtual machine software environment that could be hosted within other virtual machine software environments without also creating significant overhead that would have a negative impact on workload performance. It is clear that this was "Rocket Science" of a very high order.
So far, so good. Ravello has developed some very interesting technology that could make development of virtualized workloads for physical and cloud environments much easier.
Unfortunately, the company decided to use marketing hyperbole, "virtualization 2.0," to describe their very innovative work.
As I point out in my O'Reilly Media book "Virtualization: A Manager's Guide"
>Virtualization is a way to abstract applications and their underlying components away from the hardware supporting them and present a logical or virtual view of these re- sources. This logical view may be strikingly different from the physical view. The goal of virtualization is usually one of the following: higher levels of performance, scalability, reliability/availability, agility, or to create a unified security and management domain.
This virtual view is constructed using excess processing power, memory, storage, or network bandwidth.
Virtualization can create the artificial view that many computers are a single computing resource or that a single machine is really many individual computers. It can make a single large storage resource appear to be many smaller ones or make many smaller storage devices appear to be a single device
I laid out a model to look at all of the layers of virtualization in my 2007 commentary Sorting out the different layers of virtualization. The model includes seven layers of technology and there are many different types of products in each layer.
My first thoughts were that Ravello, like many innovative companies in the past, developed something interesting and then claimed their innovation in one narrow area of virtualization technology was so new and different that it would raise the bar for everyone, everywhere, in all areas of virtualization technology. Having been the head of marketing at a software startup years ago, I can understand the urge. One had to be careful, however, not to over-reach to a level that it obscures the message you're trying to get across.
When viewed from a neutral position, it is clear that while the company is doing something both innovative and interesting, what they're doing is in the area of virtual machine technology and not a new take on all sevel layers of virtualization technology in use today.
Furthermore, the company offers a highly edited version of the history of virtualization technology that leaves out most of the innovation, most of the different types of virtualization technology and presents a world in which their approach just naturally follows from the industry trends.
Virtualization 2.0? Not even close.