At XenSource, we believe that virtualization ought to be a ubiquitous feature of every system and pervasive in all IT operations. And so, our agenda as a company is really to make sure that virtualization shows up everywhere. That the principle that there's a hypervisor across all systems such that we can unlock much of the value in IT today.
Completely agree. This is the right way for XenSource (and VMware) to be thinking about the problem. Go big or stay home. Hypervisors everywhere provide for the best underlying management layer to break the ties that bind business processes to physical hardware today. Low-level virtualization enables this to happen. You could almost argue that this is a "winner takes all" market. Whoever owns this layer owns a very big business. VMware certainly has a significant lead today and would be the easy one to pick as the ultimate winner. I like a market with underdogs, however. This will be fun to watch!
To get to ubiquity, thinking out 5-10 years, you can almost imagine the hypervisor being shipped as part of the hardware itself. My view of how this may play out:
- Step 1 was for Intel and AMD to support hardware-assisted virtualization. Done, finally. XenSource argues that they just need a paravirtualized OS and the hardware-assisted virtualization is not required. I disagree. Customers don't want paravirtualized operating systems. They want a "standard" OS (however you define that). Also, I believe that hardware-assisted virtualization is fundamentally required to make all this play, and be secure, in the future. The good news is that this change is happening now, thanks to Intel and AMD. From a technology perspective, I like AMD's approach better, but that really doesn't matter. Check this box and let's move on.
- Step 2 is for companies like XenSource and VMware to offer hypervisors in software that provide for that common layer of OS manageability. Getting there. Xen has a different approach than VMware, today, and they argue that their approach is the only one that will work and scale downstream. That's the bet. I'm not convinced. I would agree that Xen is better technology and a better architectural approach. But, that is not what is required to win in the market. VMware ESX + hardware-assisted virtualization is "good enough".
- Step 3 is for Intel and AMD to build the hypervisor into their servers. That's how customers would like to consume this technology. It will not be "ubiquitous" without this happening. OK, this is the 5-10 year part. In the end, I think of this stuff as just another part of the BIOS (nasty word). The "hypervisor" is the next generation of what you get when you order a server from Intel and AMD. Unfortunately, the BIOS and boot environment is not one that has historically accepted such a technological change. I.e., odds are slim that this actually happen in the 5-10 year window. If the hypervisor becomes truly ubiquitous, then the value, as expected, is provided by the management tools that manipulate. AMD will lead here, and Intel will follow. VMware is better positioned today to form these partnerships, but XenSource absolutely has the opportunity to beat them. XenSource has added some key BizDev folks with the additions of Peter and Frank Artale, so it's theirs to lose.
Advantage: VMware (but too early to call).
So, lots of work for both teams. VMware gets my nod today because they are generating cash like crazy (and the related market domination that comes with that). It's very hard for XenSource to overcome their momentum, but they could do it through some key business development activities in the next 18 months. XenSource should know where the schmoozing will have to happen by now. XenSource will not win based on their technology. Let the games begin!
Tags: Startups, Xen, XenSource, VMware, hypervisor, LinuxWorld, Brian Berliner, brianberliner