VMware has made it very clear that it views virtualization as the catalyst technology which enables driving complexity and cost out of the data center, and injecting freedom, agility, and choice into the data center. VMware has also made it clear that these benefits from virtualization will only occur if virtualization itself (the hypervisor and the low level hypervisor management tools) are complemented by a layer of management tools that allow applications that are hosted on internal and external clouds to be managed very differently than they have been in the past.
Specifically, it is VMware’s observation that for each major application or class of application a silo of tools, processes and people has grown up around that application or class of applications. The plethora of different, incompatible, overlapping, non-integrated and inflexible tools and processes that have grown up around these silos have caused IT to spend too much money maintaining an unresponsive and brittle infrastructure.
VMware intends to address this issue largely by building its own end-to-end management stack on top of its hypervisor. Therefore VMware intends to become a management software company, whose management stack is justified by and powered by the benefits enabled by virtualization. This new management stack (parts of which exist today), will be built in competition with the third party software vendors who comprise the VMware ecosystem today.
VMware and Microsoft Management Stacks
There several sets of very important questions that need to be answered in order to understand the implications of VMware’s strategy.
Is Being a Management Company the Right Strategy for VMware?
Let’s compare and contrast three potential strategies for VMware:
- The vSphere Platform Strategy. This strategy would focus upon establishing a dominant market share position for the vSphere Platform (hypervisor, vCenter, and whatever else was essential for every customer to have). The platform would contain all of the API’s that are currently in the Enterprise Plus edition of vSphere, allowing third party members of the ecosystem to easily add value to the platform. The benefit of this strategy is that if successful it would create a new high share and high volume leader in the system software market (displacing Microsoft Windows in that role) – and potentially make VMware into a Microsoft like software franchise for the next 20 years (as Windows has been for the last 20 years). The problem with this strategy is that Microsoft and Red Hat are determined to bundle for free into their operating systems everything that VMware would put into the platform.
- The VMware Management Stack Strategy. This strategy (as described above) basically concedes to Microsoft and Red Hat that the lower layers of virtualization (hypervisor and basic admin tools) will be commoditized and free, and focuses VMware upon the creation of value at the management stack layers of virtualization. The benefit of this strategy is that it allows VMware to create and sell very high value management software, capitalizing upon the disruptive opportunity created by virtualization. The issues with this strategy is that it places VMware in the position of promising an end-to-end management stack across a wide range of use cases and market segments – largely in competition with members of the ecosystem who provide this functionality as point products today.
- The Have it Both Ways Strategy (also known as having your cake and eating it too). This strategy is based upon the establishment of the vSphere Platform in the same manner as described in #1 above, and VMware building management tools for some but not all use cases and market segments. It also requires VMware to be clear with its ecosystem parters where it is going to compete and where it is going to cooperate – and to actually cooperate at a sales and marketing level where it has said the door is open.
The prevailing wisdom is that VMware has to pursue either strategy #2 or #3, because to focus on #1 would be to play into Microsoft’s hands. However, to become a management company means competing with a set of vendors (see below) with whom VMware has never really competed, and selling in a way that VMware has never sold.
With Whom is VMware Actually Competing?
If in fact VMware does what it says it is going to do, and becomes all or in part a vendor of a new management stack, then VMware does not really end up with other hypervisor vendors as its primary and only competition. VMware will end up competing the following sets of vendors:
- The “Big Four” of management software. This set consists of CA, the Tivoli division of IBM, and Management Software Division of HP (comprised of the OpenView and previously acquired Mercury operations), and BMC.
- The vendor determined to make the “Big Four” into a “Big Five” EMC Ionix. It is highly ironic that VMare’s biggest competition on the management front may in fact come from its own parent company EMC, of which Ionix is a wholly owned subsidiary. Ionix will be the subject of its own set of analyses here at the Virtualization Practice, but suffice it to say that the company is acquiring the necessary pieces at a rapid rate, and intends to create a management suite that will bridge the evolution of the data center from its physical to its virtual incarnation.
- Virtualization Management focused suite vendors like VizionCore and Veeam. Both of these vendors have substantial customer and VAR traction, and loyal followings in both communities. They are also both very nimble and do a great job of staying connected with the virtualization market.
- Every software vendor at VMworld and then some. It is probably not too much of an exaggeration to say that if your product was based in software, and you were exhibiting at VMworld that you were probably on VMware’s product roadmap.
What is Microsoft’s Strategy?
It is clear that Microsoft’s strategy will be rooted in two fundamental concepts. The first will be its Windows legacy which teaches Microsoft that an inferior platform (Windows vs Unix) can easily win out due to superior ISV support. This is exactly what happened in the mid-90’s as Windows took on Unix (HP UX, IBM AIX, SCO, etc.) with a technically inferior operating system and a superior ISV ecosystem. The second is that Microsoft must make a virtue of having a less filled out enterprise virtualization management stack than VMware. This means that Microsoft, must and will be more open and supportive of the third party ecosystem than VMware. Microsoft has a history of working with third party ISV’s that add value to its platform, even if those ISV’s compete with a Microsoft “application”. Microsoft needs only to establish sufficient traction in mid-size enterprises and above in order for the third party ISV community to see this opportunity. This puts Microsoft in the position of being able to leverage the existing and rich Windows ISV community as well as potentially co-opt (or at the minimum gain the support of) the existing VMware ecosystem.
What is the VMware Ecosystem Going to Do?
Right now the vendors that target specific opportunities in the VMware environment have for the most part a one or a two year lead over the VMware offerings themselves. This is true across security, management and performance – three of the most important areas of the stack. However, every ISV who supports VMware is looking over their shoulder expecting more competition over time. Software vendors in the VMWare ecosystem will be focusing heavily upon continuing to build their differentiation with respect to VMWare, and building out a customer base in these one to two years while VMware builds out its management stack. During these two years, the ISV community will also hope that Microsoft Hyper-V establishes serious traction in production virtualization – at least in the mid-size enterprise customer accounts. As this happens, these ISV’s (those that have not done so yet) will add support for Hyper-V (and potentially other hypervisors) to their offerings and position themselves as cross-platform, objective, and fair management stacks.
What Should Enterprise Customers Do?
Perhaps the most important question is what should an enterprise customer with an existing and significant commitment to running production applications on VMware do? It is very clear that VMware vSphere has performance, scalability and manageability features today that make it appropriate as a platform to virtualize tier 2 and even tier 1 applications, and that Hyper-V does not have the same capabilities yet. However these same enterprises need to ask themselves the following question. When you committed to Windows, did you commit to buying all of the management tools for Windows from Microsoft? As you committed to various flavors of Unix/Linux over the years did you buy your management tools for those platforms from the platform vendor. When you committed to Cisco for switches and routers, did you buy all of your network management tools from Cisco?
Never before in the post mainframe history of the software business, has the vendor of a platform also been the dominant vendor of the management tools for that platform. The Virtualization Practice therefore recommends that enterprises with aggressive virtualization plans on the VMware platform continue with those plans, as there is no reasonable short term substitute for the vSphere platform. However, enterprises should draw a line somewhere above the hypervisor, and aggressively evaluate third party tools along with offerings from VMware for all layers of the stack above that line.
For example, it probably does not make any sense to look to third party vendors to replace vCenter, or vMotion. This functionality can logically be considered part of the vSphere platform. However functions like security, VM lifecycle management, configuration management, self-service provisioning, infrastructure performance management, application performance management, and transaction performance management should all be considered above the line.
Finally enterprises need to make some philosophical and architectural decisions. Should you be a multiple hypervisor shop (just as most enterprises are a multiple OS shop)? Deciding to be a multiple hypervisor shop does not mean that you put Hyper-V in production in place of vSphere today. As a matter fact it may mean that you do not slow down your vSphere deployments at all. What it does mean is that you go back to the concept of the line in the stack mentioned above and pick a place where you expect to put in cross hypervisor tools so that you can manage multiple hypervisors in the same way as they attain the required enterprise maturity. Enterprises that do decide to be multiple hypervisor shops should obviously put Hyper-V R2 into the same kinds of low hanging fruit environments (test, development, and low stress production) that were first virtualized with VMware two years ago. Having both virtualization platforms in house, along with a variety of third party management tools will facilitate choice at the platform level, and the most productive management strategy possible.