At the OpenStack Design Summit yesterday, Gordon Mangione of Citrix stated in his keynote that VMware takes almost 80% of all virtualization revenue. Regardless of how accurate this number is, it is large, and so it should not be surprising that VMware wants to protect that revenue stream.
But the market is rapidly moving up the stack. Alternate virtualization technologies are free and often come with the operating system (Xen, KVM, Microsoft) and attention is shifting to the PaaS layer. The thinking is: if there is PaaS, as a developer, why should I have to bother with virtual machines if I can simply submit my app through some API, and it magically runs on as many virtual machines as required? So what used to be virtualization/IaaS revenue becomes revenue for a combination of PaaS+IaaS, instead of IaaS revenue alone, with the perception that PaaS provides the value, while the lower-level virtualization layer is a commodity.
For many, myself included, VMware’s announcement of an open-source, Apache-licensed (i.e. do whatever you want with it, including incorporating it in commercial closed-source products) PaaS project was a surprise. But with the above market background, it makes perfect sense: by promoting a free PaaS layer that can run on any cloud, it becomes much harder for PaaS providers to argue that they are where the value is at, and thus it becomes harder for them to make money. Which means they cannot easily displace VMware’s revenue position in virtualization.
Which reminds us of Christensen’s Innovator’s Dilemma. He famously argued that in a technology stack, adjacent layers are often alternating between free and expensive. VMware’s CloudFoundry PaaS play is clearly intended to entrench a free PaaS layer and thereby protect VMware’s virtualization revenue stream.
Clever! The only question is: will it work? Isn’t the threat for them much more a PaaS software plus operations, instead of just PaaS technology? Or is that why they are also starting the hosted play cloudfoundry.com?