You gotta love provocative titles such as the one plastered on this article! But I really don’t mean the content itself to be provocative. There’s recently been a lot of discussion about how VMware will fare now that Microsoft has released Hyper-V and is committing its considerable resources and energies to win in the Hypervisor space. I just figured I’d share my perspectives with you. The timing seems to be good┬ábecause this exact question – the future of VMware – has been coming up quite a bit in customer conversations and is clearly an important┬áconsideration for companies looking to invest large amounts of time and capital in virtualization rollouts.
First off, I feel compelled to share a little bit of software industry history. Back in 1986/87, the revolutionary Intel 80386 chip came out. It brought to market several important innovations, such as virtual mode (yes, being “virtual” is not a new concept!) and access to extended memory. The 386’s virtual 8086 mode essentially allowed you to emulate a large number of concurrent 8086 processors with several performance and robustness (partitionining) optimizations. If Intel’s marketing dept could see the future, they would probably have called this feature “Virtual Cores”, but apparently the DeLorean wasn’t working properly that day!
On the software side, Multitasking was the hot systems software concept of the era. It was all about who could do the best job at running multiple applications on the same processor at the same time in a safe and efficient way. Windows 1.0 (1985), the Macintosh and several systems prior to even these seminal products showed how the multi-window + pointer paradigm could allow for intuitive task switching by non-technical users. This was a concept that clearly had arrived on the big stage and the timing was perfect. With the power of hindsight, we know the profound impact all these technologies had on computing.
However, in 1987, the product that represented the leading-edge was not a name many of you would recognize today. It outperformed nearly everything else, and took advantage of the 80386’s hardware innovations beautifully. It was a product known as DESQview-386, built by a California based company called Quarterdeck. I still remember the absolute┬áexhiliration I felt when I ran DESQview-386 and prior versions on my computer! Multiple windows, automatic visual adjustments for apps that had never been designed to run in a multitasking environment, peripheral sharing across applications and fabulous memory management via the Quarterdeck QEMM product.┬áA stellar piece of technology!┬áThe product innovation continued for a few years after DESQview-386 – DESQview/X which in many ways was too advanced for its time, was also extremely impressive.
DESQview/X in all its glory
But since I am not using DESQview’s latest version right now, and neither are you, we know it didn’t win. So what became of DESQview and Quarterdeck? Well, DESQview’s sophistication lost to Microsoft’s persistence, competitiveness and good old fashioned strategic outmaneuvering. Quarterdeck was acquired by DataStorm and then eventually by Symantec. Some piece parts of Quarterdeck products made their way into the Symantec portfolio, but by and large, it’s now only relevant for historians.
Why did I waste 7 minutes of your valuable time telling you this story? Because to me, the situation the Hypervisor industry finds itself in today is quite similar. And if there’s one thing the Cylons of Battlestar Galactica have taught us, it’s this: “It has all happened before. And it will all happen again.” VMware, like Quarterdeck, is undoubtedly the┬áinnovator in this new area of systems software (new, at least, for the PC!). Hypervisor technology is a hot category today, like multi-tasking software was in the mid-late ’80s. Hardware optimizations are being delivered to allow Virtualization to be more effective (e.g. Intel VT, AMD Pacifica) just like the real-mode enhancements to the 80386 enabled multitasking. Hypervisor technology is so important, key components of it are being baked into the processor itself. Just like multi-tasking enhancements are what drove processor design from the 80386 on.
When something becomes too important for its own good, though, strange things happen. Hypervisors are becoming commoditized because they must. They are important to the future of computing. When Windows started leveraging the channel Microsoft had established for their DOS product, Quarterdeck had no hope of competing. Even though, for a time, they continued to have the more powerful product. Ultimately, Microsoft Microsoft delivered iterative enhancements that chipped away at all the competitive differentiation Quarterdeck and others had. And while I am no Cylon, I will tell you, Hyper-V will be to the Windows channel, what Windows was to the DOS channel. Believe it.
So, with this background, what is VMware’s future? If VMware can’t figure out how to very rapidly (within 24 months), earn a substantial percentage of their revenue from non-Hypervisor products, they will be the second act (or the third? fourth?) in the DESQview saga.
Thus far, VMware’s initial reaction to the commoditization of the Hypervisor has been to get into the applications business. Specifically, apps that run on the Hypervisor. I’ll point out that I consider Virtual Center to be an exception to this, because it’s not so much a Hypervisor-enabled app as an app that enables the Hypervisor. Big difference. But other than Virtual Center, when you look at the Propero acquisition and the resulting VDM product, or the Akimbi acquisition and the resulting Virtual Lab product, you are essentially looking at examples of VMware riding roughshod over partners at a time when it requires its partners to be loyal to it. Today, dozens and dozens of ISVs that initially thought VMware was trying to be an “infrastructure company”, in a nutshell, are feeling double-crossed. It appears to them that VMware is now trying to be an applications company with offerings in multiple different areas. The same partners that contibuted to the VMware mystique by hyping ESX and the rest of the lineup, extolling on their virtues in┬ápresentations to customers, ┬áare now finding it beneficial to go the direction of Hyper-V. With little love lost.
You could ask, “Well, wouldn’t Microsoft get into virtualization applications eventually?” Maybe, but for most partners, it’s a distant possibility as compared to the competitive reality that an applications-focused VMware presents. And, while VMware lives or dies (today) on the strengths of its Hypervisor business, Microsoft can afford to not make money┬áfrom Hyper-V for at least ten years and still come out smiling. So, at the very least,┬áMicrosoft’s motivation to step into virtualization applications may be a little curtailed. Certainly not so for VMware.
For customers who see this reality emerging, large scale investments in VMware desktop infrastructure – which is a vehicle to sell a larger number of ESX licenses than server virtualization could ever manage -┬á merit a second look. For partners that are now competing with VMware in one area or the other, figuring out how to poke holes in the VMware product lineup is now an every day reality. And for partners that are lucky enough to not be competitive with VMware today, there is no guarantee that they can sleep easy. The only thing for them to do is to build support for other Hypervisors so when The Day comes, they are prepared. And lastly, for VMware itself, if it defines the new frontier as applications instead of infrastructure, it then faces the prospect of competing with a company that is 50-60 times larger than it, at a time when its own partners are having serious doubts about the “quality of the relationship” they enjoy.
So if it’s not with applications competitive to their partners, what is the product evolution and revenue diversification strategy for VMware? I’ll get to that. Another day, another time.