While some inevitably poke fun (e.g. the Mystical Quadrilateral, the Prescient Parallelogram, etc.), most any IT vendor will tell you it’s very serious business indeed.
Years of effort are spent trying to painfully nudge our particular dot upwards and to the right. Gartner doesn’t hand out free passes.
Once you get your product into a favored position (e.g. one of the leaders), the goal then shifts to create as much distance as you can between your dot and everyone else in your category.
This is the fifth year Gartner has recognized VMware as one of the leaders in the hypervisor category. Per Gartner’s rules, vendors can’t claim to be the singular leader, just one of the plural leaders.
But as I read through the Gartner analysis and criteria for judging, it felt almost historical in nature. How Gartner has historically viewed hypervisor choice — and how I think enterprise IT groups now view hypervisor choice — have clearly separated.
Not Many Surprises — On The Surface
You’d be best served by reading Gartner’s full report here.
Having stared at Gartner MQs for many years, the first thing that jumps out is the wide separation between the two leaders. Most Gartner MQs you’ll see have the leaders tightly clustered around each other, with only infinitesimal separation requiring maximum zoom to discern.
The visible implication here is that the two “leaders” are in reality not all that close — which would be my personal observation as well.
Notice that Oracle VM peeks over the horizon as a “challenger”. I don’t think Gartner could have placed their dot any lower in that quadrant.
All other hypervisors are banished to the “niche player” category.
Now, On To The Details
Right off the bat, Gartner claims ‘at least’ 70% of x86 workloads have already been virtualized. So, you might ask, what’s left to be virtualized?
One healthy component would be servers in very small settings — server closets and the like — the “small” part of SMB — where you’ll still find a lot of yet-to-be-virtualized servers.
Personally, I believe this niche is where Microsoft is seeing some success with their Hyper-V feature — these undemanding environments typically have very limited IT staff (e.g. 1-3 people) that already need to support Microsoft operating systems and tools anyway, so using the Hyper-V feature included as part of Windows makes a certain sense.
If you read through the commentary, the Gartner analysts seem to think Microsoft primarily has a sales and marketing challenge, and not product issues. Perhaps this is a result of feature-checklist comparisons. I would disagree with Gartner’s assessment — I think there are significant and meaningful differences in the respective technologies.
The people I know who use vSphere tend to routinely push it to its limits. I have yet to meet (or hear of) anyone who’s pushing Hyper-V hard — or trying anything particularly ambitious. Those that do often end up as VMware customers.
I would think this is a material difference between the two technologies.
Gartner cites their own December 2013 survey, where more than 90% of the respondents consider vSphere their primary hypervisor, and 48% named Hyper-V as their secondary hypervisor. I find this curious — what, exactly, does “secondary” mean?
Given the enormous pressure for IT organizations to standardize on common automation, technology stacks, etc. — I have to openly question what significant role — if any — a “secondary” hypervisor would play in most enterprises.
While running production on multiple hypervisors might sound useful in theory, I don't meet many people working in IT who would consider that an ideal state of affairs.
Where Gartner Might Be Missing It
In all fairness, Gartner’s perspective is on well-defined and generally-accepted historical industry categories, with x86 hypervisors being one of those. That’s what analysts do.
Back then, the major attraction was server consolidation: more logical servers on less physical hardware. That’s still somewhat important: one can still differentiate on how many logical servers can be loaded up, performance, efficiency, features, etc.
But for many, I think the focus has moved on to bigger goals. These days, I believe there are two important strategic vectors in play when it comes to hypervisor choice.
The first goal is automation — the choice of hypervisor greatly impacts the degree of automation sophistication that can be achieved, and how quickly. Server consolidation was all about capex efficiency, advanced automation is all about opex efficiency.
Yes, there are a great many automation tools available today that are positioned as hypervisor-agnostic, but — for most time-pressed enterprise IT groups — the natively-integrated automation tools (e.g. vCAC for vSphere and other hypervisors) hold great appeal.
The second strategic vector is that there’s more to the hypervisor discussion these days than familiar server virtualization. Software-defined networking (and software-defined storage) is a real thing.
Should these be considered stand-alone functions, or — ideally — seen as extensions of the server hypervisor?
The idea of taking three independently-conceived and engineered “software-defined” stacks (compute, network, storage) and integrating them into a single software-defined data center strikes me as an unnecessarily difficult (and inherently unoptimized) path.
Certainly if one wants to automate converged workflows across all three disciplines.
My belief is that — once people get their head wrapped around the desirability of these newer forms of infrastructure virtualization, there will be strong appeal in native integration vs. after-the-fact interoperability.
Not Everyone Sees It This Way
Sampling bias: I spend the majority of my time with larger and/or progressive IT organizations. My views are shaped by theirs. That being said, there are plenty of IT consumers that are neither large nor progressive. Pragmatism rules: people do what makes sense for them.
Anything beyond simple automation and UIs is probably not a priority for this crowd. They want native, intuitive and built-in workflows that work the way they do.
Similarly, software-defined networking and storage will only be appealing to them once it’s neatly baked into the hypervisor they’re using.
I think VSAN is a good example of this.
Humble Yet Purposeful
You might think the latest Gartner results would encourage VMware to boast proudly and loudly as many vendors would be tempted to. Not the case -- an obligatory matter-of-fact press release, a blog post, and not much more.
The mood is humble yet purposeful. The people I work with are incredibly demanding on themselves and their colleagues. Nothing is “good enough” for them. Their consensus is clear: we’ve only scratched the surface on what can be achieved. For most, IT infrastructure is still unnecessarily complex and inefficient.
While VMware has certainly achieved much, there is so much more that needs to be done.
Like this post? Why not subscribe via email?