VMware's Bold New Plan
VMware's challenge is to convince current and potential customers that "good enough" virtualization solutions aren't. Is it up to the task?
VMworld 2008 marked a very important moment for VMware as a company. Heading into VMworld in September, VMware was a company on the defensive and was clearly starting to feel the pressure from its competitors, namely Microsoft and Citrix. VMware has maintained a technology edge, but for the first time has had to face the "good enough" factor.
Technology that is good enough will do a particular job -- often without many of the bells and whistles of its competitors -- at a significantly lower price point. So it's possible that VMware, even with the best virtualization technology, could eventually succumb to its competitors. History has proven on numerous occasions that the best technology doesn't always win in the long run, and VMware needs to execute on a number of fronts to hold back competitors that are "good enough."
VMware CEO Paul Maritz and CTO Steve Herrod articulated the company's response to "good enough" at VMworld 2008; if you missed the keynotes, I recommend viewing them here. Maritz and Herrod disclosed an ambitious vision at the conference; let's dig in and see what they revealed.
There were too many announcements made at the conference to give them all equal attention, so instead I'll focus on the ones I think are most significant:
- Virtual Datacenter OS (VDC-OS)
- vStorage APIs
- vClient Initiative
Get Your Head in the Clouds
If you turn back the clock a couple of years, vendors were lining up to share their virtualization vision; today, the same thing is happening with cloud computing. So it's no surprise that Maritz unveiled VMware's cloud vision in his VMworld keynote. Cloud-based computing, with virtualization as a key enabler, is starting us down a path where the way we fundamentally manage IT services will be radically different. Many organizations are already far down the path to delivering IT as a service, where IT owns physical assets and departments lease virtual space on those assets. When you look beyond IT-as-a-service, the future gets even more interesting. Take VMware's vClient Initiative, for example.
vClient is all about allowing users to access their applications and data from any device -- laptop, thin client, desktop, smartphone or iPhone -- at any location. Regardless of device or location, the experience will remain relatively the same.
Of course, to make this type of transparency work, we will have to have an infrastructure that supports this level of mobility. As we transition to cloud-based computing, two cloud models will start to take shape: the public cloud and the private cloud. The public cloud will be delivered by a variety of Web service providers and allow organizations to lease physically isolated or shared infrastructures on which to run their applications.
Shared infrastructures are substantially cheaper than dedicated hardware, but many compliance considerations have kept shared infrastructures from being mainstream platforms for production applications. So until we have acceptable security standards for hosting resources in the public cloud, enterprises will manage their own private clouds as well. The private cloud will have the same look and feel as the public cloud, but will be managed internally.
VMware's vision of orchestration software automatically spinning up Web servers in the public cloud to meet unexpected demand, for example, is much more fantasy than reality for the typical enterprise, and I don't expect that to change until acceptable security standards exist for virtualization- and cloud-based computing. Most security auditors today still prefer physically isolated security zones, so there remains a significant chasm that must be crossed to go from physically isolated internal infrastructure to virtualized and shared public infrastructure. We'll get there, but it won't be tomorrow. VMsafe-enabled security appliances that will ship in 2009 are going to help, but work still remains, primarily in the standards bodies that define security compliance.
Of course, in this cloud-based IT model, there will still need to be software that touches the metal, and that's where VMware's VDC-OS comes into the mix. VMware feels that it's the ideal platform for the next-generation service-oriented IT delivery model; one which offers users secure and transparent access to the applications and data they need to do their jobs. There are a few reasons why VMware's VDC-OS and vCloud are very compelling:
- OS and application agnostic
- Limited lock-in
- Ease of deployment and use
Many of the early cloud providers offer Web-based applications that require a proprietary programming interface; so, an organization using these services must write or deploy applications specifically designed for that cloud service. VMware, on the other hand, sees a cloud that can natively support all Windows and Linux applications and operating systems, thus giving organizations greater flexibility to deploy applications to the cloud and to switch external cloud providers if they choose. To me, this architecture is the most compelling and is well positioned for uptake and long-term sustainability.
I wrote about VMware's upcoming vStorage APIs in last issue's column ("More than One Way to Skin a Disk," September/ October 2008). The vStorage APIs enable an array partner to understand which blocks on a Virtual Machine File System (VMFS) volume comprise an individual VM, allowing storage vendors to extend their granular storage-management offerings such as snapshots and replication to VMs residing on VMware's proprietary VMFS. In the past, VMware shops often had to give up some array functionality in order to obtain the benefits of VMFS. The intent of the vStorage APIs is to remove feature sacrifices in exchange for VMFS. Instead, organizations will be able to have their cake (simple virtual infrastructure storage management) and eat it too (ability to leverage the array's advanced storage-management features).
vClient is another big initiative by VMware. Prior to the arrival of Paul Maritz, I thought that VMware was moving far too slowly on the desktop. VMware's desktop vision portrays a user environment that can be accessed anywhere on any device, and not surprisingly vendors such as Citrix and Microsoft share a similar vision. So winning the desktop is perhaps VMware's greatest challenge.
Regardless of the winner, all three of these vendors are taking the desktop down a similar path -- one that includes single-instance management of desktop OSes and apps. VMware plans to offer single-instance management at the storage level using linked-cloned virtual disks, or via storage partners that offer deduplication in their arrays.
The bottom line here is that the way we will manage desktops will fundamentally change over the next five to 10 years. Many of the traditional desktop-management tools will no longer be needed, and it's this cosmic shift in desktop management that offers opportunity for vendors that wish to challenge Microsoft in the desktop space.
There's no doubt that this will be a tough battle, but given the emergence of Web 2.0-enabled client-side applications, it's possible that tomorrow's next-generation client system will look radically different than it does today. Odds remain in Microsoft's favor that it will be the vendor to bring it to us. However, VMware isn't going to concede the desktop space, and it's strategically important to strive for a best-of-breed solution that encompasses both servers and desktops. After all, Microsoft was able to take the lead in the server space primarily due to the numerous management hooks that exist between Windows client and server applications. So if VMware can't win -- or at least be a major factor in the desktop space -- there's a good chance that history will repeat itself.
Laying out a vision is the easy part, while execution always provides the greater challenge. VMware took some heat for calling itself an OS, but let's face it; the virtual infrastructure is taking some of the roles away from the traditional OS. So to me, VMware's VDC-OS branding makes perfect sense.
Who Will Go up in Smoke?
Not since Cheech and Chong has so much attention been paid to clouds. VMware intends to be the intelligence layer that manages the majority of both internal and external clouds. Maritz has laid out an ambitious goal, and Herrod has hinted at the technologies that they feel will make it all happen.
To win, VMware will need to deflect the heat from its competitors that threatens to vaporize its cloud vision, and the company will need to present a clear case as to why organizations should choose VMware products over those perceived as "good enough." Virtualization is an infrastructure technology, and like most infrastructure technologies, a virtual infrastructure platform will not be replaced on a whim. With that in mind, VMware will need to do everything in its power to accelerate virtualization adoptions on its platform, while clearly articulating reasons why "good enough" alternatives may not be as good as they appear.
Chris Wolf is VMware's CTO, Global Field and Industry.