Howard Marks has published a post I’ve been meaning do myself, but to be honest, I’m glad Howard put it out there. His is way more researched and comprehensive than mine would ever have been.
In his The True Cost Of Hyperconvergence article, Howard compares buying a new EVO:RAIL system, with building your own. Complete with the required hardware, licenses and support contracts. The result might come as a surprise to some…
One question Howard doesn’t ask though is this; What happens to the bundled VMware licenses after the initial three years? The initial cost of the EVO:RAIL includes 3 years SnS, but what happens in year 4? I guess you can buy more years of support, and extend the period easily, but I have not verified this in any way.
But this opens up for another question; What happens when you replace your EVO:RAIL after 3, 4 or 5 years? Do you have to buy a new one, complete with new licenses, even if you have paid SnS for the entire period, and what happens to the bundled VMware licenses when you replace your EVO:RAIL?
As far as I can see, the licenses follow the hardware (think Microsoft OEM licensing here), so if you decide to replace it, you have to acquire new licenses for your new hardware. At least Microsoft offers OEM licenses at a discount.
So it seems that not only do you not get discounted VMware licenses when you purchase an EVO:RAIL, you also don’t get to keep the licenses if you ever replace the hardware.
Don’t get me wrong here, I love the EVO:RAIL concept and I had really high hopes for it, but sadly I feel that VMware has missed the mark here with quite some margin. In my opinion this could have made real ripples in a lot of datacenters, and helped smaller businesses «get with the SDDC program», but with it’s current price point and licensing issues, I just don’t see it happening.
As a concept it’s really solid. As a physical form factor, it’s brilliant. As a quick delivery method for quite complex software, it’s amazing. Sadly all of this comes at a premium, a premium I’m unsure if the market is really willing to pay, at least not in it’s current iteration.