Watching VMware roll out vSphere was a bit surreal. VMware did a great job of explaining the mission of vSphere 5 (attack the next 60% – virtualize any application). VMware did a great job of positioning its cloud suite (vSphere 5 and all of the products announced along with it) as the suite of functionality needed to address that next 60% of un-virtualized workloads. In fact all of the analyst briefings that occurred before the actual launch focused upon these two elements as did the first 90% of the launch. And then…
VMware Announced New Pricing
At first this seemed like it was going to be well received. The first bit of news was encouraging – the number of vSphere Editions was going to shrink from 6 to 5, with Advanced going away, and Advanced users getting upgraded to Enterprise for no additional charge – nice move.
Then it was announced that the old core restrictions on core per CPU were going to go way. Now we had two new elements of simplicity – second nice move.
And then “Oh by the way we have a new vRAM entitlement that will become a licensed entity. We have looked at what our customers have today, and we think that this will affect only a few of the use cases at a few of our customers”.
Could This Be a Fatal Mistake?
One of the interesting things about companies that have dominant market positions like VMware is that they are rarely done in by one of their competitors. Most of the time it is a self-inflicted wound. The Motley Fool asked a very timely question, “Is VMware Killing Itself“, noting that solving problems that do not exist can be fatal, as can solving a problem in a way that creates painful unintended consequences (and you thought that only governments made these mistakes). The Fool also points out that VMware’s stock is priced to assume perfect execution in the future.
Could This Affect the Future of Virtualizing Business Critical Applications with vSphere?
The single most dangerous part of this new pricing (to VMware) is rooted in the following fact. What is left to virtualize is very different from what has been virtualized to date. If what VMware has done is change its licensing around to replace one metric (cores) with another (vRAM) in a manner that would have allowed it to get the same revenue from its existing customers to date, then VMware has totally missed the boat.
VMware made dramatic improvements to the performance and capacity of the vSphere platform with the expressed intent of focusing the platform upon the 60% of the business critical and performance critical applications that have not been virtualized yet. It is a simple fact that these applications are VERY different than the ones that have been virtualized to date. They typically generate a whole lot more network I/O and storage I/O (for which VMware made provision in vSphere 5). They also typically consume a great deal more vRAM and VMware quadrupled the amount of memory that a single VM could use from 256GB to 1000GB.
The Part That Does Not Make Sense – And That Could be Fatal
So VMware comes out with a new version of vSphere that is designed to allow the virtualization of the remaining 60% of the applications that are not virtualized yet. They specifically put features into the platform to allow it to address high memory applications – meaning that someone had to have known that the next set of applications were going to use a lot more memory that the ones that have gotten virtualized to date. And then they turn around and change their pricing to make it extra expensive (essentially a vRAM tax) to virtualize the very high memory (business critical) applications that the future success of VMware is riding upon.
This is potentially fatal mistake because the people that own those business critical applications see only risk and no benefit to their virtualization. The IT teams have been struggling to either get permission from the applications owners, or overcome objections to virtualization from these applications owners. Now those IT teams are going to have to re-plan how much it will cost to virtualize these applications, and do the proposed ROI for their management all over again. In the very best of cases this will take a bunch of virtualization projects off of the table until the financial planning is redone and the budgets are re-approved with higher license costs as a part of the deal.
In the worst of all worlds (for VMware), these projects go way entirely. Those applications either never get virtualized, or those projects get recast around competing products (Microsoft Hyper-V, Citrix Xen, Red Hat KVM).
Can This Be Fixed?
For VMware the strategically important part of this issue lies with its enterprise accounts the vast majority of whom have Enterprise License Agreements where the customer does not pay what is on the price sheet anyway. The first step that VMware needs to take is to offer an addendum to all of its existing ELA’s that allow customers to continue to roll out vSphere 5 at the same per CPU price (irrespective of vRAM) as they were were paying with vSphere 4. Doing this will allow the largest and most strategic rollouts of virtualization and private clouds to continue uninterrupted. Failing to do this will derail VMware’s progress on its most import initiatives with its most important customers.
VMware then needs to address the general issue of how vSphere 5 is priced and packaged. Given how negatively this new pricing has been received up on the vSphere Community Forums, a time out and a do-over would seem to be the right steps.