Overconfidence is a common affliction among wildly successful companies (a problem we should all be lucky enough to have!). When your stock is trading at more than 10X revenues, you’re growing at 35% YoY and you are enjoying 80% marketshare in your core market, it’s easy to see how the leadership becomes less prone to self-critique. And in this danger lurks. Just ask the good folks at Netflix.
VMware is evolving in a number of ways and much of their strategy makes great sense. Increasingly, they are becoming an IT infrastructure management company with the vSphere platform as the razor upon which many additional management “blades” can be sold. This transition is happening quickly as you can see from this chart from their recent analyst day presentation. Notice that while 74% of Q2 (Enterprise License) bookings came from “infrastructure” software (vSphere), already 12% of (ELA) bookings (light blue, far right bar) in Q2 came from “management” products, up from near zero a year earlier. So far, so good.
But they might just be assuming a bit too much as a given when it comes to their cloud management stack. For example, their pricing policies for this class of software are not per socket or per server but per VM. And if you buy into the whole stack, their own estimates (see chart) tell us that each additional VM will cost nearly twice the cost of the hypervisor. So, consolidate 10 loads onto one server and your total cost will be about 3X what the hypervisor would cost. Ouch.
Now think about this. VMware has made its business preaching the true cost-saving benefits of server consolidation. But now, by way of their cloud stack, every additional “consolidated” server load is going to rapidly escalate costs such that you can easily get upside down on the “savings”. Now in fairness, you wouldn’t necessarily adopt Site Recovery Manager (disaster recovery) for every VM and right now, SRM and vCloud Director don’t actually work together anyway. So à la carte selections may mitigate the total cost impact. But the presentation tells us the directional intention.
This kind of pricing has the appearance of great confidence that customers won’t look elsewhere for management software or that other choices won’t be viable. It also places a variable and unpredictable cost into the IT budget for cloud infrastructure long before usage-based chargeback is really feasible or practical. And so, this strategy creates quite a pricing umbrella to shelter VMware’s cloud management competitors, particularly for those that choose a scheme other than per VM.
And pricing is not the only sign of potential overconfidence. There is also no support for other hypervisors across any of VMware’s management solutions. Now with VMware’s dominance, this has a ring of good practical sense. But the message to customers is, when you buy into the VMware cloud stack, you are also deciding that you will never use other hypervisors in your cloud. Lockin strategies all but beg CIOs to evaluate other options.
There’s more. The VMware management stack doesn’t handle nonvirtualized servers (of which there will always be some)
and it doesn’t handle physical provisioning (another important part of the puzzle) while a number of the competitive products handle both do. So really, we’re talking about an immature offering. An old boss had a great slogan for such a situation: “You can get better, but you can’t pay more!”
Now the folks at VMware are smart. I think we are likely to see changes in the above in the quarters ahead. But in the meantime, companies like Abiquo, DynamicOps and Platform Computing (now IBM) and others with their infrastructure-agnostic cloud management software offerings are surely going to benefit from the current VMware cloud strategy. From an industry health standpoint, this is probably good news.