We are a top 100 websites (SAS), and we use a small licensed ESXI as a proof of concept(and a small number of free ESXI). Any though of moving our hosting infrastructure to VMware has been canceled as a result of the new license model. Under the 4.x license we were seriously considering moving a ton of machines to free/paid ESXI depending on their purpose. As you mentioned with a load balancer and redundancy, who cares if one of your web pool goes down.
With the new license, the hosting infrastructure has said emphatically we will NEVER consider VMware deployment.
Our corporate infrastructure (a completely separate network and security model from our hosting infrastructure) currently runs on 4.0 enterprise plus. We are planning to freeze on 4.0, until it end of life in 2014-2016. Perhaps a year or so from now we will start to proof Xen and Hyper-V as feasible alternatives. Honestly this make me sad, as I love the VMware product.
Assuming you are not an idiot, changing hypervisors is something done carefully and with long term testing and proof that it is reliable, so any change in market share as result of this model will take a year or two to become visible.
Many customers ignored other alternatives Xen and Hyperv-V. As any MBA knows loyal customers are price insensitive up to a point, which is the case with VMware. This is dependent on their customer loyalty, they don't mind paying more because they feel it is a fair deal. Once they feel as though they have been unfairly treated, they get mad, and reconsider price as a consideration. I can understand the thinking behind VMware's license, new processors allow higher consolidation, we are getting more value from the product, and they want to be paid for that value we receive. Customers bought on the precept that including using all the resources on their systems now and into the future. Under their current licensing model, which is pricier than the alternatives (and more feature rich) they were making money hand over fist, revenues up 30% per year…so why the change?
Looking down the road 2-years when standard 2-socket entry level servers have 20 cores, I can understand VMware's hesitancy to continue the current model. Customers will scale out and potentially need even less licenses.
Although this is EXACTLY what happened with the 55xx Intel Nehalem chipset, performance dramatically improved for virtualization (So dramatic in fact the Intel chip test team could not believe the performance results they were getting on the first 55xx prototypes and kept retesting.) even with the dramatic improvement in CPU performance, revenues still went up.
Vsphere is the engine that drives all the other growth from all their other wonderful products, and suddenly they want to put the brakes on that growth. They cannot up-sell on their platform if people don’t ante up to play on their platform. I would consider that a significant portion of their future revenue stream is from the add-ons that are licenses per VM.
I would consider the mantra, “It’s the platform, stupid.”, get customers on the platform, keep them on the platform and charge them for the value on add-ons that run on top of the platform. Don’t raise the buy-in to get onto the platform to stop them from playing at the VMware table. Make them buy new stuff on the platform, that is the future revenue.
