Home > Blogs > Virtual Reality > Monthly Archives: July 2008

Monthly Archives: July 2008

Recipe for a Catchy Headline: Top ___ Reasons Why VMware is ___ (Fill in the blanks)

Once again, another report recently came out predicting VMware’s doom in light of the formal market entry of Microsoft Hyper-V. This time, the predictions of fire and brimstone come from Clabby Analytics in the form of a paper titled “Six Reasons Microsoft’s Hyper-V Will Overtake VMware and Become the Major Player in the x86 Server Virtualization Marketplace.” This theme, VMware’s upcoming battle, seems to be a favorite topic among certain analysts, bloggers, and press nowadays as it makes for an eye-catching headline and probably drives up ad revenues for certain bloggers. But the reality is that 1) there is little substance behind these reports to back the dramatic headline, and 2) the authors’ conclusions are often based on incomplete analysis and generic “conventional wisdom” / thinking-of-the-masses.

(Quick Note: Microsoft also seems to like these types of reports as evidenced by how quickly the Clabby Analytics report showed up on microsoft.com for download. Or did Microsoft somehow have a hand to play in this report? Hmmm….)

Now, don’t get us wrong, Microsoft is a formidable competitor whenever they enter any market. No company should ever treat Microsoft’s entry lightly – that would be foolish. But here’s the reality: No one (including these pundits) really knows what will happen over the next 3-5 years in the fast changing virtualization market. There are lots of predictions flying around – everyone puts forth their hypotheses. It is then up to readers to judge them on the merits of the arguments. In judging the Clabby Analytics report on its merits, one quickly sees that the report’s arguments are fairly shallow (repeating generic claims that may or may not determine the final outcome) and the analysis is incomplete (conveniently leaves out much that is in VMware’s favor.)

So the way we see it, with the GA launch of Hyper-V, Microsoft has finally shipped a product that at least allows it to compete. No more excuses about Microsoft Virtual Server being a hosted product or Hyper-V being in beta. Let’s compete in the marketplace and let our respective companies’ ingenuity, innovation, and execution (and subsequent customer adoption) be the determiner of the outcome. At this point, a report declaring one company’s demise is premature at best, and a thinly veiled marketing tactic at worst.

Game on.

Six reasons cited by the Clabby Analytics report and how to interpret them:

  • Price
  • Packaging
  • Depth
  • Reach
  • Control/integration
  • Installed base

Reach and Installed Base Arguments

Let’s look at Reach and Installed base first.

  • For Reach, the report basically sums up the argument like this: “In short, Microsoft is significantly larger than VMware.”
  • The Installed base argument follows a similar line of logic: “Microsoft has a massive installed base, expected to cross 1 billion users by the end of 2008. By contrast, VMware claims to have 100,000 customers. Clearly Microsoft has a huge edge here…” (My note: how many of those 1 billion users are Windows Server users?)

The common message behind both of these arguments seems to be 1) Microsoft is bigger and has more resources, 2) Microsoft has a massive existing presence, and therefore, VMware doesn’t stand a chance. While the two statements about Microsoft’s size and presence are factually true, it’s kinda like saying “Bill Gates and Steve Ballmer are wealthy individuals” – factually true, but not terribly insightful. If one extends this argument to its logical conclusion, then all the Davids of the world, past present and future, might as well never bother getting out of bed to confront their respective Goliaths – the status quo will remain the status quo and the incumbents will remain the incumbents. Whether you buy into this line of reasoning or not determines whether you buy into 2 of the report’s 6 reasons – that’s one third of the report’s substance right there. (Incidentally, I’m sure that similar arguments were made when Microsoft first began its battles against IBM back in the 80s.)

Price Argument

Next, let’s take a look at the Price argument. The report shows a comparison of upfront software license acquisition costs between VMware and Microsoft. Some discrepancies and issues jump out right away.

1. Wrong Comparison
The report compares VMware Infrastructure (VI3) Enterprise Edition (VMware’s higher-end SKU) to Microsoft’s offering but leaves out the fact that Microsoft’s solution lacks core virtualization capabilities found in VI3 Enterprise Edition, such as:

  • VMware ESXi (ultra-thin hypervisor)
  • VMware VMotion (live VM migration)
  • VMware Storage VMotion (live storage migration)
  • VMware Dynamic Resource Scheduling (dynamic load balancing)
  • VMware Distributed Power Management (automated power savings)
  • VMware VMFS (distributed journaling file system built for virtualization).

A much more appropriate product comparison would have been to compare the Microsoft Hyper-V hypervisor to VMware ESXi which lists for $495 per license. And for small businesses that want a full virtual infrastructure over just a hypervisor, VMware offers the VI3 Foundation Edition and VMware SMB Acceleration Kits at very affordable price points. The report fails to mention any of these other VMware offerings.

2. VM Density Matters
Before virtualization, IT would run one application per physical server as the best practice. Therefore, for 10 hypothetical applications, people would commonly calculate the cost to run 10 servers as a proxy for the cost to run those 10 applications. The Clabby Analytics report takes this outdated approach. But with virtualization, things change dramatically – IT now runs many applications (each in its own VM) per physical server. So to figure out the cost of running those 10 applications, one must know how many VMs can run per physical server with acceptable performance – a metric called “VM density per physical host.” VMware invested in technologies to achieve very high VM density on ESX while maintain high levels of responsiveness. In fact, ESX can commonly run twice the number of VMs compared to competitors – on the same physical server. This higher density leads to less servers, licenses, power, cooling, and datacenter space. Therefore, due to VMware’s ability to achieve greater VM density per server, VMware’s cost per application, or cost per virtual machine, is much less than competing virtualization offerings. There is no mention of this important metric in the Clabby Analytics report.

One final note on Price: It’s interesting that Microsoft is pitching lower cost of upfront acquisition with Hyper-V while VMware’s focus is on total-cost-of-ownership and value. It wasn’t that long ago that Microsoft started arguing total-cost-of-ownership and value as to why people should not purchase Linux. Microsoft still has a full website dedicated to this argument. Funny how Microsoft’s position has flip-flopped when it comes to Hyper-V.

Depth Argument

Clabby Analytics argues that Depth is where Microsoft has its greatest advantage over VMware. Quote: “However, when it comes to actual management of physical and virtual servers combined, Clabby Analytics argues that Microsoft is far deeper from a management and infrastructure perspective than VMware.” Again, it is true that VMware VirtualCenter does not manage physical non-virtualized servers. Our strategy has always been to open up our management APIs so that existing systems management vendors can have deep integrations with VirtualCenter (whether it be HP, IBM, CA, BMC, Symantec, Quest, NetIQ, even Microsoft, to name a few). This way, customers can keep whatever systems management tool they’ve already purchased, and VMware integrates with it – preserving a customer’s existing management investment and best practices, and reducing disruption caused by introducing virtualization into the datacenter. With the recent expansion of Microsoft System Center to manage Linux and Unix platforms, it appears that Microsoft is really setting itself up to compete against the existing systems management vendors (like HP, IBM, CA, BMC, Symantec, Quest, NetIQ, etc.). VMware, on the other hand, partners with these vendors so the customer can leverage whatever they’ve already got for systems management.

Packaging and Control/Integration Arguments

That leaves Packaging and Control/integration as the final 2 of the 6 reasons.

On Packaging, the Clabby Analytics report states that Hyper-V’s distribution as part of Windows Server 2008 is a key advantage – it makes it easy for people to get and try Hyper-V. Yep – this tactic has always been part of Microsoft’s long-standing product strategy of leveraging Windows and distribution through PC OEMs. But the report should also have mentioned that customers can order VMware ESXi embedded from all major server OEMs. Nor does it mention how IT admins can have a new ESXi-embedded server up and running within minutes of pulling it out of the box.

Regarding Control/integration, the report states that VMware products are an “add-on” to Windows and because of this, VMware is at Microsoft’s mercy when it comes to working with Windows. First, in what way is VMware a “Windows add-on”? Is there some confusion with GSX Server here? VMware ESX is certainly no Windows add-on since it runs bare-metal on the hardware. Second, the notion that Microsoft will somehow use control of Windows to disadvantage other vendors gets into tricky territory. Isn’t that the type of behavior that got Microsoft in trouble with the DOJ and EU in the first place?

Conclusion

So judge for yourself. Do the arguments in the Clabby Analytics report make a strong case for VMware’s demise? Or is it an incomplete, one-sided analysis that basically repeats generic arguments that have been proven in the past to not be universal predictors of outcome?